From 59237fb55965ade8a53d3e9e101ca4f828880eee Mon Sep 17 00:00:00 2001 From: eust-w Date: Mon, 29 Jun 2026 18:36:35 +0800 Subject: [PATCH] =?UTF-8?q?=F0=9F=9A=80=20Add=20independent=20builtin=20op?= =?UTF-8?q?erator=20images?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- backend/operator-runner/main.py | 12 +- .../builtin/act_policy_train/operator.py | 6 +- .../builtin/act_policy_train/repository.json | 46 + .../repository.json | 45 + .../agibot_world2026_validate/repository.json | 46 + .../aloha_hdf5_validate/repository.json | 46 + .../builtin/audit_sample/repository.json | 44 + .../bop_dataset_validate/repository.json | 47 + .../bos_incremental_scan/repository.json | 44 + .../repository.json | 46 + .../calvin_episode_validate/repository.json | 46 + .../repository.json | 46 + .../coco_annotation_validate/repository.json | 45 + .../builtin/command_model_operator.py | 64 - .../builtin/conditional_route/repository.json | 44 + .../builtin/csv_to_json/repository.json | 44 + .../dataset_provenance_audit/repository.json | 46 + .../repository.json | 47 + .../repository.json | 45 + .../dataset_split_manifest/repository.json | 45 + .../dexycb_dataset_validate/repository.json | 47 + .../diffusion_policy_train/operator.py | 6 +- .../diffusion_policy_train/repository.json | 46 + .../builtin/directory_scan/repository.json | 44 + .../repository.json | 46 + .../repository.json | 45 + .../episode_boundary_validate/repository.json | 45 + .../repository.json | 47 + .../repository.json | 46 + .../graspnet_dataset_validate/repository.json | 48 + .../hand_object_pose_validate/repository.json | 47 + .../repository.json | 45 + .../builtin/hf_dataset_probe/repository.json | 44 + .../ho3d_dataset_validate/repository.json | 47 + .../builtin/human_review/repository.json | 44 + .../joint_limit_validate/repository.json | 46 + .../json_field_extract/repository.json | 44 + .../builtin/json_filter/repository.json | 44 + .../builtin/json_map_rename/repository.json | 44 + .../builtin/json_to_csv/repository.json | 44 + .../repository.json | 45 + .../lerobot_dataset_validate/repository.json | 46 + .../repository.json | 46 + .../builtin/lerobot_summary/repository.json | 44 + .../repository.json | 46 + .../repository.json | 48 + .../libero_hdf5_task_validate/repository.json | 47 + .../repository.json | 46 + .../mcap_topic_summary/repository.json | 45 + .../media_reference_validate/repository.json | 49 + .../operators/builtin/merge/repository.json | 44 + .../mjcf_actuator_validate/repository.json | 45 + .../builtin/model_inference/operator.py | 5 +- .../builtin/model_inference/repository.json | 44 + .../builtin/mongodb_write/repository.json | 44 + .../builtin/openvla_finetune/operator.py | 15 +- .../builtin/openvla_finetune/repository.json | 46 + .../builtin/openvla_inference/operator.py | 15 +- .../builtin/openvla_inference/repository.json | 46 + .../operators/builtin/package/repository.json | 44 + .../parquet_table_summary/repository.json | 45 + .../repository.json | 48 + .../pointcloud_summary/repository.json | 45 + .../builtin/real_source/repository.json | 44 + .../rgbd_frame_pair_validate/repository.json | 48 + .../rh20t_scene_validate/repository.json | 47 + .../rlbench_dataset_validate/repository.json | 46 + .../rlds_episode_summary/repository.json | 44 + .../repository.json | 46 + .../robomimic_hdf5_validate/repository.json | 46 + .../repository.json | 46 + .../roboset_hdf5_validate/repository.json | 46 + .../repository.json | 46 + .../repository.json | 46 + .../ros_joint_state_validate/repository.json | 46 + .../builtin/rosbag_summary/repository.json | 45 + .../repository.json | 47 + .../repository.json | 46 + .../repository.json | 44 + .../builtin/shell_command/repository.json | 44 + .../sim3d_asset_generate/repository.json | 45 + .../builtin/structured_model_operator.py | 383 + .../builtin/template_render/repository.json | 44 + .../repository.json | 45 + .../repository.json | 45 + .../repository.json | 45 + .../umi_zarr_dataset_validate/repository.json | 46 + .../operators/builtin/upload/repository.json | 45 + .../builtin/video_annotate/repository.json | 45 + .../builtin/video_filter/repository.json | 44 + .../builtin/video_quality/repository.json | 45 + .../builtin/video_slice/repository.json | 45 + .../builtin/video_target_clip/repository.json | 48 + .../vla_model_evaluate/repository.json | 45 + .../vla_model_inference/repository.json | 45 + .../builtin/vla_model_train/repository.json | 45 + .../repository.json | 46 + .../webdataset_shard_summary/repository.json | 45 + .../xplatform_dataset_index/repository.json | 44 + .../repository.json | 44 + .../yolo_detection_validate/repository.json | 46 + .../yolo_embodied_detection_train/operator.py | 8 +- .../repository.json | 46 + .../yolo_video_target_clip/repository.json | 49 + .../zarr_trajectory_summary/repository.json | 45 + docs/operators/ACT-策略训练.md | 28 +- docs/operators/Diffusion-Policy-训练.md | 30 +- docs/operators/OpenVLA-微调训练.md | 32 +- docs/operators/OpenVLA-推理.md | 22 +- docs/operators/YOLO-具身目标检测训练.md | 26 +- docs/operators/模型推理.md | 8 +- internal/api/builtin_operators_catalog.json | 15499 ++++++++++------ internal/api/operator_image_archive.go | 2 +- .../api/poc_operator_workflow_acceptance.go | 190 +- .../poc_operator_workflow_acceptance_test.go | 27 +- internal/api/server.go | 83 +- internal/domain/types.go | 68 +- internal/store/memory.go | 37 +- internal/store/mongo.go | 37 +- .../builtin/act_policy_train/Dockerfile | 19 + .../action_state_quality_validate/Dockerfile | 19 + .../agibot_world2026_validate/Dockerfile | 19 + .../builtin/aloha_hdf5_validate/Dockerfile | 19 + .../builtin/audit_sample/Dockerfile | 19 + .../builtin/bop_dataset_validate/Dockerfile | 19 + .../builtin/bos_incremental_scan/Dockerfile | 19 + .../Dockerfile | 19 + .../calvin_episode_validate/Dockerfile | 19 + .../camera_calibration_validate/Dockerfile | 19 + .../coco_annotation_validate/Dockerfile | 19 + .../builtin/conditional_route/Dockerfile | 19 + .../builtin/csv_to_json/Dockerfile | 19 + .../dataset_provenance_audit/Dockerfile | 19 + .../Dockerfile | 19 + .../dataset_split_leakage_validate/Dockerfile | 19 + .../builtin/dataset_split_manifest/Dockerfile | 19 + .../dexycb_dataset_validate/Dockerfile | 19 + .../builtin/diffusion_policy_train/Dockerfile | 19 + .../builtin/directory_scan/Dockerfile | 19 + .../droid_raw_episode_validate/Dockerfile | 19 + .../embodiment_metadata_validate/Dockerfile | 19 + .../episode_boundary_validate/Dockerfile | 19 + .../force_torque_sensor_validate/Dockerfile | 19 + .../Dockerfile | 19 + .../graspnet_dataset_validate/Dockerfile | 19 + .../hand_object_pose_validate/Dockerfile | 19 + .../hdf5_demonstration_summary/Dockerfile | 19 + .../builtin/hf_dataset_probe/Dockerfile | 19 + .../builtin/ho3d_dataset_validate/Dockerfile | 19 + .../builtin/human_review/Dockerfile | 19 + .../builtin/joint_limit_validate/Dockerfile | 19 + .../builtin/json_field_extract/Dockerfile | 19 + .../builtin/json_filter/Dockerfile | 19 + .../builtin/json_map_rename/Dockerfile | 19 + .../builtin/json_to_csv/Dockerfile | 19 + .../language_instruction_validate/Dockerfile | 19 + .../lerobot_dataset_validate/Dockerfile | 19 + .../Dockerfile | 19 + .../builtin/lerobot_summary/Dockerfile | 19 + .../Dockerfile | 19 + .../Dockerfile | 19 + .../libero_hdf5_task_validate/Dockerfile | 19 + .../maniskill_trajectory_validate/Dockerfile | 19 + .../builtin/mcap_topic_summary/Dockerfile | 19 + .../media_reference_validate/Dockerfile | 19 + operator-images/builtin/merge/Dockerfile | 19 + .../builtin/mjcf_actuator_validate/Dockerfile | 19 + .../builtin/model_inference/Dockerfile | 19 + .../builtin/mongodb_write/Dockerfile | 19 + .../builtin/openvla_finetune/Dockerfile | 19 + .../builtin/openvla_inference/Dockerfile | 19 + operator-images/builtin/operator-images.json | 2070 +++ operator-images/builtin/package/Dockerfile | 19 + .../builtin/parquet_table_summary/Dockerfile | 19 + .../Dockerfile | 19 + .../builtin/pointcloud_summary/Dockerfile | 19 + .../builtin/real_source/Dockerfile | 19 + .../rgbd_frame_pair_validate/Dockerfile | 19 + .../builtin/rh20t_scene_validate/Dockerfile | 19 + .../rlbench_dataset_validate/Dockerfile | 19 + .../builtin/rlds_episode_summary/Dockerfile | 19 + .../Dockerfile | 19 + .../robomimic_hdf5_validate/Dockerfile | 19 + .../robomind_hdf5_quality_validate/Dockerfile | 19 + .../builtin/roboset_hdf5_validate/Dockerfile | 19 + .../robot_model_asset_validate/Dockerfile | 19 + .../roboturk_real_hdf5_validate/Dockerfile | 19 + .../ros_joint_state_validate/Dockerfile | 19 + .../builtin/rosbag_summary/Dockerfile | 19 + .../rtx_action_schema_validate/Dockerfile | 19 + .../segmentation_mask_validate/Dockerfile | 19 + .../sensor_timestamp_sync_validate/Dockerfile | 19 + .../builtin/shell_command/Dockerfile | 19 + .../builtin/sim3d_asset_generate/Dockerfile | 19 + .../builtin/template_render/Dockerfile | 19 + .../tf_transform_tree_validate/Dockerfile | 19 + .../tfrecord_structural_validate/Dockerfile | 19 + .../trajectory_integrity_validate/Dockerfile | 19 + .../umi_zarr_dataset_validate/Dockerfile | 19 + operator-images/builtin/upload/Dockerfile | 19 + .../builtin/video_annotate/Dockerfile | 19 + .../builtin/video_filter/Dockerfile | 19 + .../builtin/video_quality/Dockerfile | 19 + .../builtin/video_slice/Dockerfile | 19 + .../builtin/video_target_clip/Dockerfile | 19 + .../builtin/vla_model_evaluate/Dockerfile | 19 + .../builtin/vla_model_inference/Dockerfile | 19 + .../builtin/vla_model_train/Dockerfile | 19 + .../vla_sample_manifest_validate/Dockerfile | 19 + .../webdataset_shard_summary/Dockerfile | 19 + .../xplatform_dataset_index/Dockerfile | 19 + .../xplatform_dataset_validate/Dockerfile | 19 + .../yolo_detection_validate/Dockerfile | 19 + .../yolo_embodied_detection_train/Dockerfile | 19 + .../builtin/yolo_video_target_clip/Dockerfile | 19 + .../zarr_trajectory_summary/Dockerfile | 19 + package.json | 2 + scripts/build-poc-images-kaniko.sh | 80 +- scripts/generate_operator_images.py | 339 + .../generate_operator_repository_manifests.py | 5 + scripts/poc-operator-workflow-acceptance.sh | 34 + scripts/poc_strict_preflight_result.py | 47 + scripts/reconcile_system_data.py | 21 +- src/lib/operators.js | 3 + src/pages/WorkflowEditor.tsx | 2 + src/types.ts | 1 + .../build_poc_images_kaniko_script_test.py | 48 + .../operator_repository_layout_test.py | 63 + ...perator_workflow_acceptance_script_test.py | 14 +- tests/backend/reconcile_system_data_test.py | 30 + 230 files changed, 19717 insertions(+), 5836 deletions(-) delete mode 100644 backend/operators/builtin/command_model_operator.py create mode 100644 backend/operators/builtin/structured_model_operator.py create mode 100644 operator-images/builtin/act_policy_train/Dockerfile create mode 100644 operator-images/builtin/action_state_quality_validate/Dockerfile create mode 100644 operator-images/builtin/agibot_world2026_validate/Dockerfile create mode 100644 operator-images/builtin/aloha_hdf5_validate/Dockerfile create mode 100644 operator-images/builtin/audit_sample/Dockerfile create mode 100644 operator-images/builtin/bop_dataset_validate/Dockerfile create mode 100644 operator-images/builtin/bos_incremental_scan/Dockerfile create mode 100644 operator-images/builtin/bridgedata_tfds_manifest_validate/Dockerfile create mode 100644 operator-images/builtin/calvin_episode_validate/Dockerfile create mode 100644 operator-images/builtin/camera_calibration_validate/Dockerfile create mode 100644 operator-images/builtin/coco_annotation_validate/Dockerfile create mode 100644 operator-images/builtin/conditional_route/Dockerfile create mode 100644 operator-images/builtin/csv_to_json/Dockerfile create mode 100644 operator-images/builtin/dataset_provenance_audit/Dockerfile create mode 100644 operator-images/builtin/dataset_schema_contract_validate/Dockerfile create mode 100644 operator-images/builtin/dataset_split_leakage_validate/Dockerfile create mode 100644 operator-images/builtin/dataset_split_manifest/Dockerfile create mode 100644 operator-images/builtin/dexycb_dataset_validate/Dockerfile create mode 100644 operator-images/builtin/diffusion_policy_train/Dockerfile create mode 100644 operator-images/builtin/directory_scan/Dockerfile create mode 100644 operator-images/builtin/droid_raw_episode_validate/Dockerfile create mode 100644 operator-images/builtin/embodiment_metadata_validate/Dockerfile create mode 100644 operator-images/builtin/episode_boundary_validate/Dockerfile create mode 100644 operator-images/builtin/force_torque_sensor_validate/Dockerfile create mode 100644 operator-images/builtin/furniturebench_pkl_demonstration_validate/Dockerfile create mode 100644 operator-images/builtin/graspnet_dataset_validate/Dockerfile create mode 100644 operator-images/builtin/hand_object_pose_validate/Dockerfile create mode 100644 operator-images/builtin/hdf5_demonstration_summary/Dockerfile create mode 100644 operator-images/builtin/hf_dataset_probe/Dockerfile create mode 100644 operator-images/builtin/ho3d_dataset_validate/Dockerfile create mode 100644 operator-images/builtin/human_review/Dockerfile create mode 100644 operator-images/builtin/joint_limit_validate/Dockerfile create mode 100644 operator-images/builtin/json_field_extract/Dockerfile create mode 100644 operator-images/builtin/json_filter/Dockerfile create mode 100644 operator-images/builtin/json_map_rename/Dockerfile create mode 100644 operator-images/builtin/json_to_csv/Dockerfile create mode 100644 operator-images/builtin/language_instruction_validate/Dockerfile create mode 100644 operator-images/builtin/lerobot_dataset_validate/Dockerfile create mode 100644 operator-images/builtin/lerobot_metadata_consistency_validate/Dockerfile create mode 100644 operator-images/builtin/lerobot_summary/Dockerfile create mode 100644 operator-images/builtin/lerobot_v3_episode_index_validate/Dockerfile create mode 100644 operator-images/builtin/lerobot_video_alignment_validate/Dockerfile create mode 100644 operator-images/builtin/libero_hdf5_task_validate/Dockerfile create mode 100644 operator-images/builtin/maniskill_trajectory_validate/Dockerfile create mode 100644 operator-images/builtin/mcap_topic_summary/Dockerfile create mode 100644 operator-images/builtin/media_reference_validate/Dockerfile create mode 100644 operator-images/builtin/merge/Dockerfile create mode 100644 operator-images/builtin/mjcf_actuator_validate/Dockerfile create mode 100644 operator-images/builtin/model_inference/Dockerfile create mode 100644 operator-images/builtin/mongodb_write/Dockerfile create mode 100644 operator-images/builtin/openvla_finetune/Dockerfile create mode 100644 operator-images/builtin/openvla_inference/Dockerfile create mode 100644 operator-images/builtin/operator-images.json create mode 100644 operator-images/builtin/package/Dockerfile create mode 100644 operator-images/builtin/parquet_table_summary/Dockerfile create mode 100644 operator-images/builtin/partnet_mobility_asset_validate/Dockerfile create mode 100644 operator-images/builtin/pointcloud_summary/Dockerfile create mode 100644 operator-images/builtin/real_source/Dockerfile create mode 100644 operator-images/builtin/rgbd_frame_pair_validate/Dockerfile create mode 100644 operator-images/builtin/rh20t_scene_validate/Dockerfile create mode 100644 operator-images/builtin/rlbench_dataset_validate/Dockerfile create mode 100644 operator-images/builtin/rlds_episode_summary/Dockerfile create mode 100644 operator-images/builtin/robocasa_hdf5_demonstration_validate/Dockerfile create mode 100644 operator-images/builtin/robomimic_hdf5_validate/Dockerfile create mode 100644 operator-images/builtin/robomind_hdf5_quality_validate/Dockerfile create mode 100644 operator-images/builtin/roboset_hdf5_validate/Dockerfile create mode 100644 operator-images/builtin/robot_model_asset_validate/Dockerfile create mode 100644 operator-images/builtin/roboturk_real_hdf5_validate/Dockerfile create mode 100644 operator-images/builtin/ros_joint_state_validate/Dockerfile create mode 100644 operator-images/builtin/rosbag_summary/Dockerfile create mode 100644 operator-images/builtin/rtx_action_schema_validate/Dockerfile create mode 100644 operator-images/builtin/segmentation_mask_validate/Dockerfile create mode 100644 operator-images/builtin/sensor_timestamp_sync_validate/Dockerfile create mode 100644 operator-images/builtin/shell_command/Dockerfile create mode 100644 operator-images/builtin/sim3d_asset_generate/Dockerfile create mode 100644 operator-images/builtin/template_render/Dockerfile create mode 100644 operator-images/builtin/tf_transform_tree_validate/Dockerfile create mode 100644 operator-images/builtin/tfrecord_structural_validate/Dockerfile create mode 100644 operator-images/builtin/trajectory_integrity_validate/Dockerfile create mode 100644 operator-images/builtin/umi_zarr_dataset_validate/Dockerfile create mode 100644 operator-images/builtin/upload/Dockerfile create mode 100644 operator-images/builtin/video_annotate/Dockerfile create mode 100644 operator-images/builtin/video_filter/Dockerfile create mode 100644 operator-images/builtin/video_quality/Dockerfile create mode 100644 operator-images/builtin/video_slice/Dockerfile create mode 100644 operator-images/builtin/video_target_clip/Dockerfile create mode 100644 operator-images/builtin/vla_model_evaluate/Dockerfile create mode 100644 operator-images/builtin/vla_model_inference/Dockerfile create mode 100644 operator-images/builtin/vla_model_train/Dockerfile create mode 100644 operator-images/builtin/vla_sample_manifest_validate/Dockerfile create mode 100644 operator-images/builtin/webdataset_shard_summary/Dockerfile create mode 100644 operator-images/builtin/xplatform_dataset_index/Dockerfile create mode 100644 operator-images/builtin/xplatform_dataset_validate/Dockerfile create mode 100644 operator-images/builtin/yolo_detection_validate/Dockerfile create mode 100644 operator-images/builtin/yolo_embodied_detection_train/Dockerfile create mode 100644 operator-images/builtin/yolo_video_target_clip/Dockerfile create mode 100644 operator-images/builtin/zarr_trajectory_summary/Dockerfile create mode 100644 scripts/generate_operator_images.py diff --git a/backend/operator-runner/main.py b/backend/operator-runner/main.py index 7c066019..ed610582 100644 --- a/backend/operator-runner/main.py +++ b/backend/operator-runner/main.py @@ -5216,6 +5216,16 @@ def select_handler(node_label: str) -> str: return "generic" +def runtime_handler_override() -> str: + handler = os.getenv("DROBOTICFLOW_OPERATOR_HANDLER", "").strip() + if not handler: + return "" + if handler not in BUILTIN_HANDLERS: + known = ", ".join(sorted(BUILTIN_HANDLERS)) + raise RuntimeError(f"DROBOTICFLOW_OPERATOR_HANDLER={handler!r} is not registered; known handlers: {known}") + return handler + + def main() -> int: run_id = os.getenv("RUN_ID", "").strip() workflow_id = os.getenv("WORKFLOW_ID", "").strip() @@ -5248,7 +5258,7 @@ def main() -> int: append_log("INFO", "Executing explicit command") artifacts = execute_command(run_id, node_id, output_dir, node_data, command) else: - handler = select_handler(node_label) + handler = runtime_handler_override() or select_handler(node_label) append_log("INFO", f"Selected built-in handler: {handler}") builtin_handler = BUILTIN_HANDLERS.get(handler, BUILTIN_HANDLERS["generic"]) if handler == "generic": diff --git a/backend/operators/builtin/act_policy_train/operator.py b/backend/operators/builtin/act_policy_train/operator.py index f51935b4..1478c8c4 100644 --- a/backend/operators/builtin/act_policy_train/operator.py +++ b/backend/operators/builtin/act_policy_train/operator.py @@ -2,7 +2,7 @@ from __future__ import annotations from typing import Any -from backend.operators.builtin.command_model_operator import HelperMap, run_command_model_operator +from backend.operators.builtin.structured_model_operator import HelperMap, run_policy_training def run( @@ -13,7 +13,7 @@ def run( inputs: list[dict[str, Any]], helpers: HelperMap, ) -> list[dict[str, Any]]: - return run_command_model_operator( + return run_policy_training( run_id, node_id, output_dir, @@ -22,6 +22,4 @@ def run( helpers, operator_name="ACT 策略训练", model_family="ACT", - recommended_gpu="1x NVIDIA A800 80GB", - recommended_gpu_memory_gb=80, ) diff --git a/backend/operators/builtin/act_policy_train/repository.json b/backend/operators/builtin/act_policy_train/repository.json index 508cc371..d0fc9b88 100644 --- a/backend/operators/builtin/act_policy_train/repository.json +++ b/backend/operators/builtin/act_policy_train/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "AI", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-act-policy-train:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-act-policy-train@sha256:d24b5417b2eab1abf5bb07a1d66d38306ea5545226d3f2a3e34bb5aee184f4f2", "metadata": "operator.json", "name": "ACT 策略训练", "operatorId": "operator-builtin-act-policy-train", @@ -18,6 +25,45 @@ "preferredSlug": "act_policy_train", "pythonImport": "backend.operators.builtin.act_policy_train.operator:run", "repositoryId": "builtin/act_policy_train", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": true, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_act_policy_train", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-act-policy-train", + "operatorSlug": "act_policy_train", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 1, + "recommendedGpuType": "NVIDIA A800 80GB", + "requiresGpu": true + }, + "runtimeDependencies": [ + "real-act-training-command", + "trajectory-demonstration-files", + "gpu-a800" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "real-act-training-command", "trajectory-demonstration-files", diff --git a/backend/operators/builtin/action_state_quality_validate/repository.json b/backend/operators/builtin/action_state_quality_validate/repository.json index 076fb8bb..08f1975e 100644 --- a/backend/operators/builtin/action_state_quality_validate/repository.json +++ b/backend/operators/builtin/action_state_quality_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-action-state-quality-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-action-state-quality-validate@sha256:96055934c67458386848e774725ea4328cb0b1fe157e77a23e8586659e7b538d", "metadata": "operator.json", "name": "动作状态数值质量校验", "operatorId": "operator-builtin-action-state-quality-validate", @@ -18,6 +25,44 @@ "preferredSlug": "action_state_quality_validate", "pythonImport": "backend.operators.builtin.action_state_quality_validate.operator:run", "repositoryId": "builtin/action_state_quality_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_action_state_quality_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-action-state-quality-validate", + "operatorSlug": "action_state_quality_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-jsonl-parquet-files", + "pyarrow" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "json-jsonl-parquet-files", "pyarrow" diff --git a/backend/operators/builtin/agibot_world2026_validate/repository.json b/backend/operators/builtin/agibot_world2026_validate/repository.json index 31cb68e7..d00df9f0 100644 --- a/backend/operators/builtin/agibot_world2026_validate/repository.json +++ b/backend/operators/builtin/agibot_world2026_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-agibot-world2026-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-agibot-world2026-validate@sha256:622499759d3ae4bb0bbcbd51069fb2525510f41d3a42a474e30b7d21d9df41b8", "metadata": "operator.json", "name": "AgiBotWorld2026 数据集校验", "operatorId": "operator-builtin-agibot-world2026-validate", @@ -18,6 +25,45 @@ "preferredSlug": "agibot_world2026_validate", "pythonImport": "backend.operators.builtin.agibot_world2026_validate.operator:run", "repositoryId": "builtin/agibot_world2026_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_agibot_world2026_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-agibot-world2026-validate", + "operatorSlug": "agibot_world2026_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "agibotworld2026-meta-data-video-files", + "pyarrow", + "parquet-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "agibotworld2026-meta-data-video-files", "pyarrow", diff --git a/backend/operators/builtin/aloha_hdf5_validate/repository.json b/backend/operators/builtin/aloha_hdf5_validate/repository.json index 7667b5da..778347cb 100644 --- a/backend/operators/builtin/aloha_hdf5_validate/repository.json +++ b/backend/operators/builtin/aloha_hdf5_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-aloha-hdf5-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-aloha-hdf5-validate@sha256:e90db43b9bf900fec24f36546724121bab2a329f9ab7b48f579e84a53eb045a8", "metadata": "operator.json", "name": "ALOHA HDF5 示教校验", "operatorId": "operator-builtin-aloha-hdf5-validate", @@ -18,6 +25,45 @@ "preferredSlug": "aloha_hdf5_validate", "pythonImport": "backend.operators.builtin.aloha_hdf5_validate.operator:run", "repositoryId": "builtin/aloha_hdf5_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_aloha_hdf5_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-aloha-hdf5-validate", + "operatorSlug": "aloha_hdf5_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "aloha-act-hdf5-files", + "h5py", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "aloha-act-hdf5-files", "h5py", diff --git a/backend/operators/builtin/audit_sample/repository.json b/backend/operators/builtin/audit_sample/repository.json index d14a5608..321694ad 100644 --- a/backend/operators/builtin/audit_sample/repository.json +++ b/backend/operators/builtin/audit_sample/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "QA", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-audit-sample:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-audit-sample@sha256:2a89baaae19e6c2c33980f41921020043888dc48db23cbc69c7de6f4670c9fc2", "metadata": "operator.json", "name": "抽检采样", "operatorId": "operator-builtin-audit-sample", @@ -18,6 +25,43 @@ "preferredSlug": "audit_sample", "pythonImport": "backend.operators.builtin.audit_sample.operator:run", "repositoryId": "builtin/audit_sample", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_audit_sampling", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-audit-sample", + "operatorSlug": "audit_sample", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-artifacts" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "json-artifacts" ], diff --git a/backend/operators/builtin/bop_dataset_validate/repository.json b/backend/operators/builtin/bop_dataset_validate/repository.json index 0cc5878e..15fb57e1 100644 --- a/backend/operators/builtin/bop_dataset_validate/repository.json +++ b/backend/operators/builtin/bop_dataset_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-bop-dataset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-bop-dataset-validate@sha256:6b99f50e0777e02df933fb196666f6bace7c251fcdfedf1630844525905d5c1b", "metadata": "operator.json", "name": "BOP 6D姿态数据集校验", "operatorId": "operator-builtin-bop-dataset-validate", @@ -18,6 +25,46 @@ "preferredSlug": "bop_dataset_validate", "pythonImport": "backend.operators.builtin.bop_dataset_validate.operator:run", "repositoryId": "builtin/bop_dataset_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_bop_dataset_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-bop-dataset-validate", + "operatorSlug": "bop_dataset_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "bop-scenewise-dataset-files", + "json-files", + "rgb-depth-mask-image-files", + "ply-model-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "bop-scenewise-dataset-files", "json-files", diff --git a/backend/operators/builtin/bos_incremental_scan/repository.json b/backend/operators/builtin/bos_incremental_scan/repository.json index ba1b4143..3cd7144e 100644 --- a/backend/operators/builtin/bos_incremental_scan/repository.json +++ b/backend/operators/builtin/bos_incremental_scan/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "IO", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-bos-incremental-scan:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-bos-incremental-scan@sha256:9077ae0f63cdf0fe76ce363ea08b84f071b2fc3c0ef516bd797a8d63eba91249", "metadata": "operator.json", "name": "BOS 增量扫描", "operatorId": "operator-builtin-bos-incremental-scan", @@ -18,6 +25,43 @@ "preferredSlug": "bos_incremental_scan", "pythonImport": "backend.operators.builtin.bos_incremental_scan.operator:run", "repositoryId": "builtin/bos_incremental_scan", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_source", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-bos-incremental-scan", + "operatorSlug": "bos_incremental_scan", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "baidu-bos" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "baidu-bos" ], diff --git a/backend/operators/builtin/bridgedata_tfds_manifest_validate/repository.json b/backend/operators/builtin/bridgedata_tfds_manifest_validate/repository.json index bc2aa80c..800133a1 100644 --- a/backend/operators/builtin/bridgedata_tfds_manifest_validate/repository.json +++ b/backend/operators/builtin/bridgedata_tfds_manifest_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-bridgedata-tfds-manifest-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-bridgedata-tfds-manifest-validate@sha256:18c89efc04d3ca65154c07851231c16953900016178d6183041850e6e7f7e2a0", "metadata": "operator.json", "name": "BridgeData TFDS-RLDS清单校验", "operatorId": "operator-builtin-bridgedata-tfds-manifest-validate", @@ -18,6 +25,45 @@ "preferredSlug": "bridgedata_tfds_manifest_validate", "pythonImport": "backend.operators.builtin.bridgedata_tfds_manifest_validate.operator:run", "repositoryId": "builtin/bridgedata_tfds_manifest_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_bridgedata_tfds_manifest_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-bridgedata-tfds-manifest-validate", + "operatorSlug": "bridgedata_tfds_manifest_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "bridgedata-tfds-rlds-directory", + "tfrecord-files", + "crc32c-structural-validation" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "bridgedata-tfds-rlds-directory", "tfrecord-files", diff --git a/backend/operators/builtin/calvin_episode_validate/repository.json b/backend/operators/builtin/calvin_episode_validate/repository.json index 731eb0db..1e26104e 100644 --- a/backend/operators/builtin/calvin_episode_validate/repository.json +++ b/backend/operators/builtin/calvin_episode_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-calvin-episode-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-calvin-episode-validate@sha256:bc90a289bb258dc0f74fd1a33d2baa5ca53182e78eed23c42106205325ae9a48", "metadata": "operator.json", "name": "CALVIN Episode 校验", "operatorId": "operator-builtin-calvin-episode-validate", @@ -18,6 +25,45 @@ "preferredSlug": "calvin_episode_validate", "pythonImport": "backend.operators.builtin.calvin_episode_validate.operator:run", "repositoryId": "builtin/calvin_episode_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_calvin_episode_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-calvin-episode-validate", + "operatorSlug": "calvin_episode_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "calvin-episode-npz-files", + "numpy", + "language-annotation-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "calvin-episode-npz-files", "numpy", diff --git a/backend/operators/builtin/camera_calibration_validate/repository.json b/backend/operators/builtin/camera_calibration_validate/repository.json index 81f7c333..743866d7 100644 --- a/backend/operators/builtin/camera_calibration_validate/repository.json +++ b/backend/operators/builtin/camera_calibration_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-camera-calibration-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-camera-calibration-validate@sha256:b43c83210cdd287d8234ff4e137586b1e5d671945ef72308246e9cceb8d7cae7", "metadata": "operator.json", "name": "相机标定校验", "operatorId": "operator-builtin-camera-calibration-validate", @@ -18,6 +25,45 @@ "preferredSlug": "camera_calibration_validate", "pythonImport": "backend.operators.builtin.camera_calibration_validate.operator:run", "repositoryId": "builtin/camera_calibration_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_camera_calibration_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-camera-calibration-validate", + "operatorSlug": "camera_calibration_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "camera-calibration-files", + "json-files", + "optional-yaml-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "camera-calibration-files", "json-files", diff --git a/backend/operators/builtin/coco_annotation_validate/repository.json b/backend/operators/builtin/coco_annotation_validate/repository.json index 9c62bc75..1129cc6b 100644 --- a/backend/operators/builtin/coco_annotation_validate/repository.json +++ b/backend/operators/builtin/coco_annotation_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "DATASET", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-coco-annotation-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-coco-annotation-validate@sha256:538e5ffdc881f6f93ae55b84c216b57954bbf662a1d1a34588ed4981eb8ab57d", "metadata": "operator.json", "name": "COCO 标注校验", "operatorId": "operator-builtin-coco-annotation-validate", @@ -18,6 +25,44 @@ "preferredSlug": "coco_annotation_validate", "pythonImport": "backend.operators.builtin.coco_annotation_validate.operator:run", "repositoryId": "builtin/coco_annotation_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_coco_annotation_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-coco-annotation-validate", + "operatorSlug": "coco_annotation_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "coco-json", + "image-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "coco-json", "image-files" diff --git a/backend/operators/builtin/command_model_operator.py b/backend/operators/builtin/command_model_operator.py deleted file mode 100644 index 922db4ac..00000000 --- a/backend/operators/builtin/command_model_operator.py +++ /dev/null @@ -1,64 +0,0 @@ -from __future__ import annotations - -import json -import os -from typing import Any, Callable - - -HelperMap = dict[str, Callable[..., Any]] - - -def run_command_model_operator( - run_id: str, - node_id: str, - output_dir: Any, - node_data: dict[str, Any], - inputs: list[dict[str, Any]], - helpers: HelperMap, - *, - operator_name: str, - model_family: str, - recommended_gpu: str, - recommended_gpu_memory_gb: int, -) -> list[dict[str, Any]]: - params = helpers["hyperparams"](node_data) - command = str(params.get("COMMAND") or params.get("SCRIPT") or "").strip() - if not command: - acceptance_mode = str(params.get("POC_ACCEPTANCE_MODE", "")).strip().lower() in {"true", "1", "yes", "y", "on"} - if acceptance_mode: - payload = { - "operator": operator_name, - "modelFamily": model_family, - "recommendedGPU": recommended_gpu, - "recommendedGPUMemoryGB": recommended_gpu_memory_gb, - "requestedGPU": os.getenv("GPU_LIMIT", ""), - "inputArtifactCount": len(inputs), - "acceptanceKind": "command-backed-gpu-model-operator", - "requiresRealCommandForProduction": True, - } - return [ - helpers["write_json_artifact"]( - run_id, - node_id, - output_dir, - "model-operator-acceptance.json", - payload, - output=helpers["output_binding"](node_data), - ) - ] - raise RuntimeError( - f"{operator_name} 需要配置真实训练/推理 COMMAND 或 SCRIPT;" - f"推荐在 A800 上请求 {recommended_gpu} 张 GPU,显存建议不少于 {recommended_gpu_memory_gb}GB。" - ) - - metadata = { - "operator": operator_name, - "modelFamily": model_family, - "recommendedGPU": recommended_gpu, - "recommendedGPUMemoryGB": recommended_gpu_memory_gb, - "requestedGPU": os.getenv("GPU_LIMIT", ""), - "inputArtifactCount": len(inputs), - "requiresRealCommand": True, - } - os.environ["DROBOTICFLOW_MODEL_OPERATOR_CONTEXT"] = json.dumps(metadata, ensure_ascii=False, sort_keys=True) - return helpers["execute_command"](run_id, node_id, output_dir, node_data, command) diff --git a/backend/operators/builtin/conditional_route/repository.json b/backend/operators/builtin/conditional_route/repository.json index d5dddfc9..12468546 100644 --- a/backend/operators/builtin/conditional_route/repository.json +++ b/backend/operators/builtin/conditional_route/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "LOGIC", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-conditional-route:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-conditional-route@sha256:0de386aacb81c0658bac316e50689d3fa56bc7403865ce3237262e1dda0ff5a4", "metadata": "operator.json", "name": "条件分流", "operatorId": "operator-builtin-conditional-route", @@ -18,6 +25,43 @@ "preferredSlug": "conditional_route", "pythonImport": "backend.operators.builtin.conditional_route.operator:run", "repositoryId": "builtin/conditional_route", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_condition_router", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-conditional-route", + "operatorSlug": "conditional_route", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-artifacts" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "json-artifacts" ], diff --git a/backend/operators/builtin/csv_to_json/repository.json b/backend/operators/builtin/csv_to_json/repository.json index 5abe6ebb..b2f14b7d 100644 --- a/backend/operators/builtin/csv_to_json/repository.json +++ b/backend/operators/builtin/csv_to_json/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "DATA", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-csv-to-json:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-csv-to-json@sha256:9137bbc331c43d783ac20b3316c606612d96c1e725c95bfdac02f6de6d163c7f", "metadata": "operator.json", "name": "CSV 转 JSON", "operatorId": "operator-builtin-csv-to-json", @@ -18,6 +25,43 @@ "preferredSlug": "csv_to_json", "pythonImport": "backend.operators.builtin.csv_to_json.operator:run", "repositoryId": "builtin/csv_to_json", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_csv_to_json", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-csv-to-json", + "operatorSlug": "csv_to_json", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "csv-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "csv-files" ], diff --git a/backend/operators/builtin/dataset_provenance_audit/repository.json b/backend/operators/builtin/dataset_provenance_audit/repository.json index 38db9b0f..3fb59a86 100644 --- a/backend/operators/builtin/dataset_provenance_audit/repository.json +++ b/backend/operators/builtin/dataset_provenance_audit/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "DATASET", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dataset-provenance-audit:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dataset-provenance-audit@sha256:641ef1697d6e61773dc587c9142a50b920738107974f7ffbacf2512302aee98f", "metadata": "operator.json", "name": "数据集来源许可审计", "operatorId": "operator-builtin-dataset-provenance-audit", @@ -18,6 +25,45 @@ "preferredSlug": "dataset_provenance_audit", "pythonImport": "backend.operators.builtin.dataset_provenance_audit.operator:run", "repositoryId": "builtin/dataset_provenance_audit", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_dataset_provenance_audit", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-dataset-provenance-audit", + "operatorSlug": "dataset_provenance_audit", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "dataset-card-files", + "license-files", + "pyyaml" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "dataset-card-files", "license-files", diff --git a/backend/operators/builtin/dataset_schema_contract_validate/repository.json b/backend/operators/builtin/dataset_schema_contract_validate/repository.json index f3f388d9..a1f2e4a2 100644 --- a/backend/operators/builtin/dataset_schema_contract_validate/repository.json +++ b/backend/operators/builtin/dataset_schema_contract_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "DATASET", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dataset-schema-contract-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dataset-schema-contract-validate@sha256:9dfcb8cf18e09069e14db67c8649352071fcdf44b1d537465d38aff532f5ba48", "metadata": "operator.json", "name": "数据集Schema合约校验", "operatorId": "operator-builtin-dataset-schema-contract-validate", @@ -18,6 +25,46 @@ "preferredSlug": "dataset_schema_contract_validate", "pythonImport": "backend.operators.builtin.dataset_schema_contract_validate.operator:run", "repositoryId": "builtin/dataset_schema_contract_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_dataset_schema_contract_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-dataset-schema-contract-validate", + "operatorSlug": "dataset_schema_contract_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "csv-json-jsonl-parquet-hdf5-npz-files", + "pyarrow", + "h5py", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "csv-json-jsonl-parquet-hdf5-npz-files", "pyarrow", diff --git a/backend/operators/builtin/dataset_split_leakage_validate/repository.json b/backend/operators/builtin/dataset_split_leakage_validate/repository.json index 889c72a5..1df60335 100644 --- a/backend/operators/builtin/dataset_split_leakage_validate/repository.json +++ b/backend/operators/builtin/dataset_split_leakage_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "DATASET", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dataset-split-leakage-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dataset-split-leakage-validate@sha256:2dde8d42ac5001e4727b6061c54970c136f7588c1bde04fa7b5e3473e49aaac4", "metadata": "operator.json", "name": "数据集切分泄漏校验", "operatorId": "operator-builtin-dataset-split-leakage-validate", @@ -18,6 +25,44 @@ "preferredSlug": "dataset_split_leakage_validate", "pythonImport": "backend.operators.builtin.dataset_split_leakage_validate.operator:run", "repositoryId": "builtin/dataset_split_leakage_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_dataset_split_leakage_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-dataset-split-leakage-validate", + "operatorSlug": "dataset_split_leakage_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "split-manifest-csv-json-jsonl-parquet-files", + "pyarrow" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "split-manifest-csv-json-jsonl-parquet-files", "pyarrow" diff --git a/backend/operators/builtin/dataset_split_manifest/repository.json b/backend/operators/builtin/dataset_split_manifest/repository.json index da496817..d96b8be8 100644 --- a/backend/operators/builtin/dataset_split_manifest/repository.json +++ b/backend/operators/builtin/dataset_split_manifest/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "DATASET", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dataset-split-manifest:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dataset-split-manifest@sha256:b2bcd44ccaa4b3c2b9a22d4f10a5d8101a0e19e249e693a501e1e506887fe1da", "metadata": "operator.json", "name": "训练切分清单生成", "operatorId": "operator-builtin-dataset-split-manifest", @@ -18,6 +25,44 @@ "preferredSlug": "dataset_split_manifest", "pythonImport": "backend.operators.builtin.dataset_split_manifest.operator:run", "repositoryId": "builtin/dataset_split_manifest", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_dataset_split_manifest", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-dataset-split-manifest", + "operatorSlug": "dataset_split_manifest", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-artifacts", + "csv-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "json-artifacts", "csv-files" diff --git a/backend/operators/builtin/dexycb_dataset_validate/repository.json b/backend/operators/builtin/dexycb_dataset_validate/repository.json index 66c592c7..84e6d3f4 100644 --- a/backend/operators/builtin/dexycb_dataset_validate/repository.json +++ b/backend/operators/builtin/dexycb_dataset_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dexycb-dataset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dexycb-dataset-validate@sha256:b412bc8d84fa31ff1233f13c0758156013ec03e9cd1092e9871b8bb8b4b9e6f1", "metadata": "operator.json", "name": "DexYCB 数据集校验", "operatorId": "operator-builtin-dexycb-dataset-validate", @@ -18,6 +25,46 @@ "preferredSlug": "dexycb_dataset_validate", "pythonImport": "backend.operators.builtin.dexycb_dataset_validate.operator:run", "repositoryId": "builtin/dexycb_dataset_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_dexycb_dataset_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-dexycb-dataset-validate", + "operatorSlug": "dexycb_dataset_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "dexycb-dataset-files", + "numpy", + "npz-label-files", + "rgb-depth-image-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "dexycb-dataset-files", "numpy", diff --git a/backend/operators/builtin/diffusion_policy_train/operator.py b/backend/operators/builtin/diffusion_policy_train/operator.py index 013ab84a..426e30e2 100644 --- a/backend/operators/builtin/diffusion_policy_train/operator.py +++ b/backend/operators/builtin/diffusion_policy_train/operator.py @@ -2,7 +2,7 @@ from __future__ import annotations from typing import Any -from backend.operators.builtin.command_model_operator import HelperMap, run_command_model_operator +from backend.operators.builtin.structured_model_operator import HelperMap, run_policy_training def run( @@ -13,7 +13,7 @@ def run( inputs: list[dict[str, Any]], helpers: HelperMap, ) -> list[dict[str, Any]]: - return run_command_model_operator( + return run_policy_training( run_id, node_id, output_dir, @@ -22,6 +22,4 @@ def run( helpers, operator_name="Diffusion Policy 训练", model_family="Diffusion Policy", - recommended_gpu="1x NVIDIA A800 80GB", - recommended_gpu_memory_gb=80, ) diff --git a/backend/operators/builtin/diffusion_policy_train/repository.json b/backend/operators/builtin/diffusion_policy_train/repository.json index 76f6dc8c..cacacf7d 100644 --- a/backend/operators/builtin/diffusion_policy_train/repository.json +++ b/backend/operators/builtin/diffusion_policy_train/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "AI", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-diffusion-policy-train:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-diffusion-policy-train@sha256:78d74d404ed7e98c19518e28315a897313c8583c6f6b0eb191e36eabcc8a2f33", "metadata": "operator.json", "name": "Diffusion Policy 训练", "operatorId": "operator-builtin-diffusion-policy-train", @@ -18,6 +25,45 @@ "preferredSlug": "diffusion_policy_train", "pythonImport": "backend.operators.builtin.diffusion_policy_train.operator:run", "repositoryId": "builtin/diffusion_policy_train", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": true, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_diffusion_policy_train", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-diffusion-policy-train", + "operatorSlug": "diffusion_policy_train", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 1, + "recommendedGpuType": "NVIDIA A800 80GB", + "requiresGpu": true + }, + "runtimeDependencies": [ + "diffusion-policy-training-command", + "zarr-or-lerobot-trajectory-files", + "gpu-a800" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "diffusion-policy-training-command", "zarr-or-lerobot-trajectory-files", diff --git a/backend/operators/builtin/directory_scan/repository.json b/backend/operators/builtin/directory_scan/repository.json index 4885693f..1e85f19b 100644 --- a/backend/operators/builtin/directory_scan/repository.json +++ b/backend/operators/builtin/directory_scan/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "IO", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-directory-scan:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-directory-scan@sha256:f591c68e5a9f63ed1203347dedfddfe7f9f8786266d69af1fbdab74bc4eece9f", "metadata": "operator.json", "name": "目录扫描", "operatorId": "operator-builtin-directory-scan", @@ -18,6 +25,43 @@ "preferredSlug": "directory_scan", "pythonImport": "backend.operators.builtin.directory_scan.operator:run", "repositoryId": "builtin/directory_scan", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_directory_scan", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-directory-scan", + "operatorSlug": "directory_scan", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "filesystem" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "filesystem" ], diff --git a/backend/operators/builtin/droid_raw_episode_validate/repository.json b/backend/operators/builtin/droid_raw_episode_validate/repository.json index 508074ff..7ef3d6fc 100644 --- a/backend/operators/builtin/droid_raw_episode_validate/repository.json +++ b/backend/operators/builtin/droid_raw_episode_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-droid-raw-episode-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-droid-raw-episode-validate@sha256:eea450070b6b1fa4fbd80c9f1e476eacba25683791b0b40320639c65cc5fa418", "metadata": "operator.json", "name": "DROID Raw Episode校验", "operatorId": "operator-builtin-droid-raw-episode-validate", @@ -18,6 +25,45 @@ "preferredSlug": "droid_raw_episode_validate", "pythonImport": "backend.operators.builtin.droid_raw_episode_validate.operator:run", "repositoryId": "builtin/droid_raw_episode_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_droid_raw_episode_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-droid-raw-episode-validate", + "operatorSlug": "droid_raw_episode_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "droid-raw-episode-hdf5-metadata-video-files", + "h5py", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "droid-raw-episode-hdf5-metadata-video-files", "h5py", diff --git a/backend/operators/builtin/embodiment_metadata_validate/repository.json b/backend/operators/builtin/embodiment_metadata_validate/repository.json index 97ba1654..4575fbb6 100644 --- a/backend/operators/builtin/embodiment_metadata_validate/repository.json +++ b/backend/operators/builtin/embodiment_metadata_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-embodiment-metadata-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-embodiment-metadata-validate@sha256:47df5e78b0fd62bf046f8340b7af5d4cef3bc38efaa2b10a3a7ee028adc62ebc", "metadata": "operator.json", "name": "多机器人Embodiment元信息校验", "operatorId": "operator-builtin-embodiment-metadata-validate", @@ -18,6 +25,44 @@ "preferredSlug": "embodiment_metadata_validate", "pythonImport": "backend.operators.builtin.embodiment_metadata_validate.operator:run", "repositoryId": "builtin/embodiment_metadata_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_embodiment_metadata_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-embodiment-metadata-validate", + "operatorSlug": "embodiment_metadata_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "csv-json-jsonl-parquet-metadata-files", + "pyarrow" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "csv-json-jsonl-parquet-metadata-files", "pyarrow" diff --git a/backend/operators/builtin/episode_boundary_validate/repository.json b/backend/operators/builtin/episode_boundary_validate/repository.json index 2ab45f02..ce7c6848 100644 --- a/backend/operators/builtin/episode_boundary_validate/repository.json +++ b/backend/operators/builtin/episode_boundary_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-episode-boundary-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-episode-boundary-validate@sha256:9e7782f3fe626f858138fd066ca2911d02cc4bfe6ab933065b4ebfedaff544ba", "metadata": "operator.json", "name": "Episode 边界校验", "operatorId": "operator-builtin-episode-boundary-validate", @@ -18,6 +25,44 @@ "preferredSlug": "episode_boundary_validate", "pythonImport": "backend.operators.builtin.episode_boundary_validate.operator:run", "repositoryId": "builtin/episode_boundary_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_episode_boundary_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-episode-boundary-validate", + "operatorSlug": "episode_boundary_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-jsonl-parquet-files", + "pyarrow" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "json-jsonl-parquet-files", "pyarrow" diff --git a/backend/operators/builtin/force_torque_sensor_validate/repository.json b/backend/operators/builtin/force_torque_sensor_validate/repository.json index 00733f29..9986e6ab 100644 --- a/backend/operators/builtin/force_torque_sensor_validate/repository.json +++ b/backend/operators/builtin/force_torque_sensor_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-force-torque-sensor-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-force-torque-sensor-validate@sha256:d7e4558fd8fad892f328af10f21eaa8ca3f8122fd7fe209184b99c5f51a7a22a", "metadata": "operator.json", "name": "力扭矩传感器数据校验", "operatorId": "operator-builtin-force-torque-sensor-validate", @@ -18,6 +25,46 @@ "preferredSlug": "force_torque_sensor_validate", "pythonImport": "backend.operators.builtin.force_torque_sensor_validate.operator:run", "repositoryId": "builtin/force_torque_sensor_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_force_torque_sensor_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-force-torque-sensor-validate", + "operatorSlug": "force_torque_sensor_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "csv-json-jsonl-parquet-hdf5-force-torque-files", + "pyarrow", + "h5py", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "csv-json-jsonl-parquet-hdf5-force-torque-files", "pyarrow", diff --git a/backend/operators/builtin/furniturebench_pkl_demonstration_validate/repository.json b/backend/operators/builtin/furniturebench_pkl_demonstration_validate/repository.json index 4ed6c2a9..1c7f6670 100644 --- a/backend/operators/builtin/furniturebench_pkl_demonstration_validate/repository.json +++ b/backend/operators/builtin/furniturebench_pkl_demonstration_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-furniturebench-pkl-demonstration-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-furniturebench-pkl-demonstration-validate@sha256:6633bd68e09a6a82997e46bd6f02a0a3b27f77ba40aab0094239cd11317f7c8c", "metadata": "operator.json", "name": "FurnitureBench PKL示教校验", "operatorId": "operator-builtin-furniturebench-pkl-demonstration-validate", @@ -18,6 +25,45 @@ "preferredSlug": "furniturebench_pkl_demonstration_validate", "pythonImport": "backend.operators.builtin.furniturebench_pkl_demonstration_validate.operator:run", "repositoryId": "builtin/furniturebench_pkl_demonstration_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_furniturebench_pkl_demonstration_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-furniturebench-pkl-demonstration-validate", + "operatorSlug": "furniturebench_pkl_demonstration_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "furniturebench-pkl-files", + "pickle", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "furniturebench-pkl-files", "pickle", diff --git a/backend/operators/builtin/graspnet_dataset_validate/repository.json b/backend/operators/builtin/graspnet_dataset_validate/repository.json index 5c1f1ccc..db16ae01 100644 --- a/backend/operators/builtin/graspnet_dataset_validate/repository.json +++ b/backend/operators/builtin/graspnet_dataset_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-graspnet-dataset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-graspnet-dataset-validate@sha256:c4ec92cc4e87e5df975377608f02ba54de0570d0235dd36840d4a311875c060a", "metadata": "operator.json", "name": "GraspNet 抓取数据集校验", "operatorId": "operator-builtin-graspnet-dataset-validate", @@ -18,6 +25,47 @@ "preferredSlug": "graspnet_dataset_validate", "pythonImport": "backend.operators.builtin.graspnet_dataset_validate.operator:run", "repositoryId": "builtin/graspnet_dataset_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_graspnet_dataset_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-graspnet-dataset-validate", + "operatorSlug": "graspnet_dataset_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "graspnet-1billion-dataset-files", + "numpy", + "npz-npy-files", + "rgb-depth-mask-png-files", + "xml-annotations" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "graspnet-1billion-dataset-files", "numpy", diff --git a/backend/operators/builtin/hand_object_pose_validate/repository.json b/backend/operators/builtin/hand_object_pose_validate/repository.json index 7d6e3e6f..345aa7be 100644 --- a/backend/operators/builtin/hand_object_pose_validate/repository.json +++ b/backend/operators/builtin/hand_object_pose_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-hand-object-pose-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-hand-object-pose-validate@sha256:f91c6c745f098d7a592c3738b4e9bbb4a2be1604605da5ccd539ef90b010deed", "metadata": "operator.json", "name": "手-物体姿态标注校验", "operatorId": "operator-builtin-hand-object-pose-validate", @@ -18,6 +25,46 @@ "preferredSlug": "hand_object_pose_validate", "pythonImport": "backend.operators.builtin.hand_object_pose_validate.operator:run", "repositoryId": "builtin/hand_object_pose_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_hand_object_pose_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-hand-object-pose-validate", + "operatorSlug": "hand_object_pose_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-jsonl-csv-parquet-files", + "image-files", + "mask-files", + "pyarrow" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "json-jsonl-csv-parquet-files", "image-files", diff --git a/backend/operators/builtin/hdf5_demonstration_summary/repository.json b/backend/operators/builtin/hdf5_demonstration_summary/repository.json index eb64a0d6..c1477b78 100644 --- a/backend/operators/builtin/hdf5_demonstration_summary/repository.json +++ b/backend/operators/builtin/hdf5_demonstration_summary/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-hdf5-demonstration-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-hdf5-demonstration-summary@sha256:3d82e565a006c7c300d45ebfffaa1098d1d6b3e8bcdffd1b8a361d0768b7e08f", "metadata": "operator.json", "name": "HDF5 示教数据概览", "operatorId": "operator-builtin-hdf5-demonstration-summary", @@ -18,6 +25,44 @@ "preferredSlug": "hdf5_demonstration_summary", "pythonImport": "backend.operators.builtin.hdf5_demonstration_summary.operator:run", "repositoryId": "builtin/hdf5_demonstration_summary", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_hdf5_demonstration_summary", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-hdf5-demonstration-summary", + "operatorSlug": "hdf5_demonstration_summary", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "hdf5-files", + "h5py" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "hdf5-files", "h5py" diff --git a/backend/operators/builtin/hf_dataset_probe/repository.json b/backend/operators/builtin/hf_dataset_probe/repository.json index 38fafa80..0c48e43b 100644 --- a/backend/operators/builtin/hf_dataset_probe/repository.json +++ b/backend/operators/builtin/hf_dataset_probe/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "DATASET", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-hf-dataset-probe:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-hf-dataset-probe@sha256:8ec0202d73059352be727b28c12daf9a6bdfd77e7e9ea3332d658efde66654c4", "metadata": "operator.json", "name": "HF 数据集探测", "operatorId": "operator-builtin-hf-dataset-probe", @@ -18,6 +25,43 @@ "preferredSlug": "hf_dataset_probe", "pythonImport": "backend.operators.builtin.hf_dataset_probe.operator:run", "repositoryId": "builtin/hf_dataset_probe", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_hf_dataset_probe", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-hf-dataset-probe", + "operatorSlug": "hf_dataset_probe", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "huggingface-dataset-viewer-api" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "huggingface-dataset-viewer-api" ], diff --git a/backend/operators/builtin/ho3d_dataset_validate/repository.json b/backend/operators/builtin/ho3d_dataset_validate/repository.json index 3266cd65..4fb0b787 100644 --- a/backend/operators/builtin/ho3d_dataset_validate/repository.json +++ b/backend/operators/builtin/ho3d_dataset_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-ho3d-dataset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-ho3d-dataset-validate@sha256:9edf87bceb97b38907f6270835c2b9fcd6dd347d800efaf41a6d747eddd8af8c", "metadata": "operator.json", "name": "HO3D/H2O3D 数据集校验", "operatorId": "operator-builtin-ho3d-dataset-validate", @@ -18,6 +25,46 @@ "preferredSlug": "ho3d_dataset_validate", "pythonImport": "backend.operators.builtin.ho3d_dataset_validate.operator:run", "repositoryId": "builtin/ho3d_dataset_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_ho3d_dataset_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-ho3d-dataset-validate", + "operatorSlug": "ho3d_dataset_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "ho3d-h2o3d-dataset-files", + "pickle", + "numpy", + "rgb-depth-image-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "ho3d-h2o3d-dataset-files", "pickle", diff --git a/backend/operators/builtin/human_review/repository.json b/backend/operators/builtin/human_review/repository.json index 94204e02..a16409d6 100644 --- a/backend/operators/builtin/human_review/repository.json +++ b/backend/operators/builtin/human_review/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "HUMAN", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-human-review:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-human-review@sha256:8e789d73f7933102ce12364c2a382affaa16639491393abfdd3092bc3090215d", "metadata": "operator.json", "name": "人工审核", "operatorId": "operator-builtin-human-review", @@ -18,6 +25,43 @@ "preferredSlug": "human_review", "pythonImport": "backend.operators.builtin.human_review.operator:run", "repositoryId": "builtin/human_review", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_human_review", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-human-review", + "operatorSlug": "human_review", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-artifacts" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "json-artifacts" ], diff --git a/backend/operators/builtin/joint_limit_validate/repository.json b/backend/operators/builtin/joint_limit_validate/repository.json index b615f089..2e6b5da5 100644 --- a/backend/operators/builtin/joint_limit_validate/repository.json +++ b/backend/operators/builtin/joint_limit_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-joint-limit-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-joint-limit-validate@sha256:b4edf3d011f7234d3479635a6aecf33c695674827487d4c8578cb02446785b82", "metadata": "operator.json", "name": "URDF 关节限位校验", "operatorId": "operator-builtin-joint-limit-validate", @@ -18,6 +25,45 @@ "preferredSlug": "joint_limit_validate", "pythonImport": "backend.operators.builtin.joint_limit_validate.operator:run", "repositoryId": "builtin/joint_limit_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_joint_limit_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-joint-limit-validate", + "operatorSlug": "joint_limit_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "urdf-files", + "json-jsonl-csv-parquet-files", + "pyarrow" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "urdf-files", "json-jsonl-csv-parquet-files", diff --git a/backend/operators/builtin/json_field_extract/repository.json b/backend/operators/builtin/json_field_extract/repository.json index d67e9b9f..19ab8b84 100644 --- a/backend/operators/builtin/json_field_extract/repository.json +++ b/backend/operators/builtin/json_field_extract/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "DATA", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-json-field-extract:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-json-field-extract@sha256:f7b6df195863eef04846a3372f02cb5cac149052496976c65891d7cb3f844945", "metadata": "operator.json", "name": "JSON 字段提取", "operatorId": "operator-builtin-json-field-extract", @@ -18,6 +25,43 @@ "preferredSlug": "json_field_extract", "pythonImport": "backend.operators.builtin.json_field_extract.operator:run", "repositoryId": "builtin/json_field_extract", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_json_extract", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-json-field-extract", + "operatorSlug": "json_field_extract", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-artifacts" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "json-artifacts" ], diff --git a/backend/operators/builtin/json_filter/repository.json b/backend/operators/builtin/json_filter/repository.json index e4229cfd..3e3b662e 100644 --- a/backend/operators/builtin/json_filter/repository.json +++ b/backend/operators/builtin/json_filter/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "DATA", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-json-filter:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-json-filter@sha256:824f00ea594090923c52d78d0b649abef1b47d37e0d8f1211734c39c66943bbb", "metadata": "operator.json", "name": "JSON 过滤", "operatorId": "operator-builtin-json-filter", @@ -18,6 +25,43 @@ "preferredSlug": "json_filter", "pythonImport": "backend.operators.builtin.json_filter.operator:run", "repositoryId": "builtin/json_filter", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_json_filter", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-json-filter", + "operatorSlug": "json_filter", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-artifacts" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "json-artifacts" ], diff --git a/backend/operators/builtin/json_map_rename/repository.json b/backend/operators/builtin/json_map_rename/repository.json index 643fe315..e6277b98 100644 --- a/backend/operators/builtin/json_map_rename/repository.json +++ b/backend/operators/builtin/json_map_rename/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "DATA", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-json-map-rename:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-json-map-rename@sha256:53a425b6fabacbf148a17aaf411ce6762c9c5dcd9f4bdf2802edd9f12fc18c72", "metadata": "operator.json", "name": "JSON 映射重命名", "operatorId": "operator-builtin-json-map-rename", @@ -18,6 +25,43 @@ "preferredSlug": "json_map_rename", "pythonImport": "backend.operators.builtin.json_map_rename.operator:run", "repositoryId": "builtin/json_map_rename", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_json_map", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-json-map-rename", + "operatorSlug": "json_map_rename", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-artifacts" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "json-artifacts" ], diff --git a/backend/operators/builtin/json_to_csv/repository.json b/backend/operators/builtin/json_to_csv/repository.json index 79b76b79..ee5a2e0d 100644 --- a/backend/operators/builtin/json_to_csv/repository.json +++ b/backend/operators/builtin/json_to_csv/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "DATA", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-json-to-csv:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-json-to-csv@sha256:17d52fd86f00c899e777859a4eef6fb4e16496cbdadbf085ec51c5e79b8bb724", "metadata": "operator.json", "name": "JSON 转 CSV", "operatorId": "operator-builtin-json-to-csv", @@ -18,6 +25,43 @@ "preferredSlug": "json_to_csv", "pythonImport": "backend.operators.builtin.json_to_csv.operator:run", "repositoryId": "builtin/json_to_csv", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_json_to_csv", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-json-to-csv", + "operatorSlug": "json_to_csv", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-artifacts" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "json-artifacts" ], diff --git a/backend/operators/builtin/language_instruction_validate/repository.json b/backend/operators/builtin/language_instruction_validate/repository.json index 483ae634..6cbfbd1f 100644 --- a/backend/operators/builtin/language_instruction_validate/repository.json +++ b/backend/operators/builtin/language_instruction_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-language-instruction-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-language-instruction-validate@sha256:302e6fee95108b952abdbd080e18641c08717983a0fa6b2efdcf387ba25e2249", "metadata": "operator.json", "name": "语言指令校验", "operatorId": "operator-builtin-language-instruction-validate", @@ -18,6 +25,44 @@ "preferredSlug": "language_instruction_validate", "pythonImport": "backend.operators.builtin.language_instruction_validate.operator:run", "repositoryId": "builtin/language_instruction_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_language_instruction_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-language-instruction-validate", + "operatorSlug": "language_instruction_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-jsonl-csv-parquet-files", + "pyarrow" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "json-jsonl-csv-parquet-files", "pyarrow" diff --git a/backend/operators/builtin/lerobot_dataset_validate/repository.json b/backend/operators/builtin/lerobot_dataset_validate/repository.json index 12a47b97..ed231c4a 100644 --- a/backend/operators/builtin/lerobot_dataset_validate/repository.json +++ b/backend/operators/builtin/lerobot_dataset_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-dataset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-dataset-validate@sha256:f2471c2ea75c05059fd995667b337681f8f6ff3a8947c5b960f03fae93a2d087", "metadata": "operator.json", "name": "LeRobot 数据集校验", "operatorId": "operator-builtin-lerobot-dataset-validate", @@ -18,6 +25,45 @@ "preferredSlug": "lerobot_dataset_validate", "pythonImport": "backend.operators.builtin.lerobot_dataset_validate.operator:run", "repositoryId": "builtin/lerobot_dataset_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_lerobot_dataset_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-lerobot-dataset-validate", + "operatorSlug": "lerobot_dataset_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "lerobot-dataset-files", + "pyarrow", + "parquet-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "lerobot-dataset-files", "pyarrow", diff --git a/backend/operators/builtin/lerobot_metadata_consistency_validate/repository.json b/backend/operators/builtin/lerobot_metadata_consistency_validate/repository.json index a5c62157..f7f0f874 100644 --- a/backend/operators/builtin/lerobot_metadata_consistency_validate/repository.json +++ b/backend/operators/builtin/lerobot_metadata_consistency_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-metadata-consistency-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-metadata-consistency-validate@sha256:a58daedea8b3996999c73020a68a07f75c1bdbbcdb6189f5d8183c322a0d42c6", "metadata": "operator.json", "name": "LeRobot 元数据一致性校验", "operatorId": "operator-builtin-lerobot-metadata-consistency-validate", @@ -18,6 +25,45 @@ "preferredSlug": "lerobot_metadata_consistency_validate", "pythonImport": "backend.operators.builtin.lerobot_metadata_consistency_validate.operator:run", "repositoryId": "builtin/lerobot_metadata_consistency_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_lerobot_metadata_consistency_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-lerobot-metadata-consistency-validate", + "operatorSlug": "lerobot_metadata_consistency_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "lerobot-v3-meta-files", + "pyarrow", + "parquet-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "lerobot-v3-meta-files", "pyarrow", diff --git a/backend/operators/builtin/lerobot_summary/repository.json b/backend/operators/builtin/lerobot_summary/repository.json index 15b43688..bddc8a4e 100644 --- a/backend/operators/builtin/lerobot_summary/repository.json +++ b/backend/operators/builtin/lerobot_summary/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-summary@sha256:c8e1878dced01da3da3a068cd53a395b03061b27f3e52e20cef2faedf4081dbf", "metadata": "operator.json", "name": "LeRobot 数据集概览", "operatorId": "operator-builtin-lerobot-summary", @@ -18,6 +25,43 @@ "preferredSlug": "lerobot_summary", "pythonImport": "backend.operators.builtin.lerobot_summary.operator:run", "repositoryId": "builtin/lerobot_summary", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_lerobot_summary", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-lerobot-summary", + "operatorSlug": "lerobot_summary", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "lerobot-dataset-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "lerobot-dataset-files" ], diff --git a/backend/operators/builtin/lerobot_v3_episode_index_validate/repository.json b/backend/operators/builtin/lerobot_v3_episode_index_validate/repository.json index 2b301de3..c82a971c 100644 --- a/backend/operators/builtin/lerobot_v3_episode_index_validate/repository.json +++ b/backend/operators/builtin/lerobot_v3_episode_index_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-v3-episode-index-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-v3-episode-index-validate@sha256:204d8dc1e1530c8096c7aca6817ce41b12ed2c70fcd0d31fe2346427b0833456", "metadata": "operator.json", "name": "LeRobot v3 Episode索引校验", "operatorId": "operator-builtin-lerobot-v3-episode-index-validate", @@ -18,6 +25,45 @@ "preferredSlug": "lerobot_v3_episode_index_validate", "pythonImport": "backend.operators.builtin.lerobot_v3_episode_index_validate.operator:run", "repositoryId": "builtin/lerobot_v3_episode_index_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_lerobot_v3_episode_index_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-lerobot-v3-episode-index-validate", + "operatorSlug": "lerobot_v3_episode_index_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "lerobot-v3-meta-data-files", + "pyarrow", + "parquet-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "lerobot-v3-meta-data-files", "pyarrow", diff --git a/backend/operators/builtin/lerobot_video_alignment_validate/repository.json b/backend/operators/builtin/lerobot_video_alignment_validate/repository.json index 3aeab064..b1105f7c 100644 --- a/backend/operators/builtin/lerobot_video_alignment_validate/repository.json +++ b/backend/operators/builtin/lerobot_video_alignment_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-video-alignment-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-video-alignment-validate@sha256:51022c000aed86bda84f3f33453984ed85620b87ea2487ed451864abcef7282a", "metadata": "operator.json", "name": "LeRobot 视频对齐校验", "operatorId": "operator-builtin-lerobot-video-alignment-validate", @@ -18,6 +25,47 @@ "preferredSlug": "lerobot_video_alignment_validate", "pythonImport": "backend.operators.builtin.lerobot_video_alignment_validate.operator:run", "repositoryId": "builtin/lerobot_video_alignment_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_lerobot_video_alignment_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-lerobot-video-alignment-validate", + "operatorSlug": "lerobot_video_alignment_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "lerobot-v3-data-files", + "pyarrow", + "parquet-files", + "video-files", + "opencv-python-headless" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "lerobot-v3-data-files", "pyarrow", diff --git a/backend/operators/builtin/libero_hdf5_task_validate/repository.json b/backend/operators/builtin/libero_hdf5_task_validate/repository.json index 35915dcf..0bca0962 100644 --- a/backend/operators/builtin/libero_hdf5_task_validate/repository.json +++ b/backend/operators/builtin/libero_hdf5_task_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-libero-hdf5-task-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-libero-hdf5-task-validate@sha256:68b8f8d5b203a0c644d8cec121c6d802ec55cf89ac0fa4aa87eb471b7076e18d", "metadata": "operator.json", "name": "LIBERO HDF5任务校验", "operatorId": "operator-builtin-libero-hdf5-task-validate", @@ -18,6 +25,46 @@ "preferredSlug": "libero_hdf5_task_validate", "pythonImport": "backend.operators.builtin.libero_hdf5_task_validate.operator:run", "repositoryId": "builtin/libero_hdf5_task_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_libero_hdf5_task_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-libero-hdf5-task-validate", + "operatorSlug": "libero_hdf5_task_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "libero-hdf5-demo-files", + "libero-bddl-files", + "h5py", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "libero-hdf5-demo-files", "libero-bddl-files", diff --git a/backend/operators/builtin/maniskill_trajectory_validate/repository.json b/backend/operators/builtin/maniskill_trajectory_validate/repository.json index dda19e5e..a0bbc453 100644 --- a/backend/operators/builtin/maniskill_trajectory_validate/repository.json +++ b/backend/operators/builtin/maniskill_trajectory_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-maniskill-trajectory-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-maniskill-trajectory-validate@sha256:d18459e85b43df78ed7df591c2ecfe86122ecb927f429cf65b465c5096edf453", "metadata": "operator.json", "name": "ManiSkill 轨迹校验", "operatorId": "operator-builtin-maniskill-trajectory-validate", @@ -18,6 +25,45 @@ "preferredSlug": "maniskill_trajectory_validate", "pythonImport": "backend.operators.builtin.maniskill_trajectory_validate.operator:run", "repositoryId": "builtin/maniskill_trajectory_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_maniskill_trajectory_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-maniskill-trajectory-validate", + "operatorSlug": "maniskill_trajectory_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "maniskill-hdf5-json-files", + "h5py", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "maniskill-hdf5-json-files", "h5py", diff --git a/backend/operators/builtin/mcap_topic_summary/repository.json b/backend/operators/builtin/mcap_topic_summary/repository.json index a9483a51..903a79d6 100644 --- a/backend/operators/builtin/mcap_topic_summary/repository.json +++ b/backend/operators/builtin/mcap_topic_summary/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-mcap-topic-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-mcap-topic-summary@sha256:421e060f3d74ac067b895d83a4c7478c43feca5a75d91623859b5a1d867a5ecc", "metadata": "operator.json", "name": "MCAP 主题概览", "operatorId": "operator-builtin-mcap-topic-summary", @@ -18,6 +25,44 @@ "preferredSlug": "mcap_topic_summary", "pythonImport": "backend.operators.builtin.mcap_topic_summary.operator:run", "repositoryId": "builtin/mcap_topic_summary", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_mcap_topic_summary", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-mcap-topic-summary", + "operatorSlug": "mcap_topic_summary", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "mcap-files", + "mcap-python" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "mcap-files", "mcap-python" diff --git a/backend/operators/builtin/media_reference_validate/repository.json b/backend/operators/builtin/media_reference_validate/repository.json index ec98d75d..6e468e07 100644 --- a/backend/operators/builtin/media_reference_validate/repository.json +++ b/backend/operators/builtin/media_reference_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "DATASET", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-media-reference-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-media-reference-validate@sha256:60bdd6320deaf66576927bf26b25809911ea98210e9c2daa0665fe9846157104", "metadata": "operator.json", "name": "媒体引用可读性校验", "operatorId": "operator-builtin-media-reference-validate", @@ -18,6 +25,48 @@ "preferredSlug": "media_reference_validate", "pythonImport": "backend.operators.builtin.media_reference_validate.operator:run", "repositoryId": "builtin/media_reference_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_media_reference_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-media-reference-validate", + "operatorSlug": "media_reference_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "csv-json-jsonl-parquet-manifests", + "image-video-depth-pointcloud-files", + "opencv-python-headless", + "numpy", + "pyarrow", + "ffprobe" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "csv-json-jsonl-parquet-manifests", "image-video-depth-pointcloud-files", diff --git a/backend/operators/builtin/merge/repository.json b/backend/operators/builtin/merge/repository.json index bbdba536..4b423be8 100644 --- a/backend/operators/builtin/merge/repository.json +++ b/backend/operators/builtin/merge/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "MATH", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-merge:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-merge@sha256:43dd9f250d94e9dcc2af193aa03a602a3f27ce4d3b119f5063760638cd1db0d9", "metadata": "operator.json", "name": "数据合并", "operatorId": "operator-builtin-merge", @@ -18,6 +25,43 @@ "preferredSlug": "merge", "pythonImport": "backend.operators.builtin.merge.operator:run", "repositoryId": "builtin/merge", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_merge", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-merge", + "operatorSlug": "merge", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-artifacts" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "json-artifacts" ], diff --git a/backend/operators/builtin/mjcf_actuator_validate/repository.json b/backend/operators/builtin/mjcf_actuator_validate/repository.json index 40a4645a..3da2908b 100644 --- a/backend/operators/builtin/mjcf_actuator_validate/repository.json +++ b/backend/operators/builtin/mjcf_actuator_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-mjcf-actuator-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-mjcf-actuator-validate@sha256:adc49f2a4cd9a27f2e74c74852c8a74eb33233e98973e80507346260e2abed8c", "metadata": "operator.json", "name": "MJCF 执行器校验", "operatorId": "operator-builtin-mjcf-actuator-validate", @@ -18,6 +25,44 @@ "preferredSlug": "mjcf_actuator_validate", "pythonImport": "backend.operators.builtin.mjcf_actuator_validate.operator:run", "repositoryId": "builtin/mjcf_actuator_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_mjcf_actuator_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-mjcf-actuator-validate", + "operatorSlug": "mjcf_actuator_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "mjcf-files", + "xml-parser" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "mjcf-files", "xml-parser" diff --git a/backend/operators/builtin/model_inference/operator.py b/backend/operators/builtin/model_inference/operator.py index 26e41882..07e12aaa 100644 --- a/backend/operators/builtin/model_inference/operator.py +++ b/backend/operators/builtin/model_inference/operator.py @@ -44,7 +44,7 @@ def load_inference_model(params: dict[str, str], helpers: HelperMap) -> tuple[di if isinstance(payload, dict): return payload, f"MODEL_PATH:{pathlib.Path(model_path)}" raise RuntimeError("MODEL_PATH must point to a JSON object model") - raise RuntimeError("模型推理需要 MODEL_JSON、MODEL_PATH、COMMAND 或 SCRIPT") + raise RuntimeError("模型推理需要 MODEL_JSON 或 MODEL_PATH;自定义命令请使用 Shell 命令算子") def numeric_feature_value(item: Any, feature: str, helpers: HelperMap) -> float: @@ -162,9 +162,6 @@ def run( helpers: HelperMap, ) -> list[dict[str, Any]]: params = helpers["hyperparams"](node_data) - command = params.get("COMMAND") or params.get("SCRIPT") - if command: - return helpers["execute_command"](run_id, node_id, output_dir, node_data, command) model, model_source = load_inference_model(params, helpers) items = model_input_items(inputs, params, helpers) predictions = [ diff --git a/backend/operators/builtin/model_inference/repository.json b/backend/operators/builtin/model_inference/repository.json index a01dc3ff..bed74575 100644 --- a/backend/operators/builtin/model_inference/repository.json +++ b/backend/operators/builtin/model_inference/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "AI", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-model-inference:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-model-inference@sha256:a31cf2777cda7b213f4b3fa7e004ed5ddcd31fe5f3503c5cbb039be1f6bfaaa4", "metadata": "operator.json", "name": "模型推理", "operatorId": "operator-builtin-model-inference", @@ -18,6 +25,43 @@ "preferredSlug": "model_inference", "pythonImport": "backend.operators.builtin.model_inference.operator:run", "repositoryId": "builtin/model_inference", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_model_inference", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-model-inference", + "operatorSlug": "model_inference", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-model-or-command" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "json-model-or-command" ], diff --git a/backend/operators/builtin/mongodb_write/repository.json b/backend/operators/builtin/mongodb_write/repository.json index 326eaca6..18a51071 100644 --- a/backend/operators/builtin/mongodb_write/repository.json +++ b/backend/operators/builtin/mongodb_write/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "DATA", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-mongodb-write:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-mongodb-write@sha256:36677374caf4ee70072c0c6ec565498565d2b667c8fb238c9ebb81bf6ecee876", "metadata": "operator.json", "name": "MongoDB 写入", "operatorId": "operator-builtin-mongodb-write", @@ -18,6 +25,43 @@ "preferredSlug": "mongodb_write", "pythonImport": "backend.operators.builtin.mongodb_write.operator:run", "repositoryId": "builtin/mongodb_write", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_mongodb_write", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-mongodb-write", + "operatorSlug": "mongodb_write", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "mongodb" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "mongodb" ], diff --git a/backend/operators/builtin/openvla_finetune/operator.py b/backend/operators/builtin/openvla_finetune/operator.py index b2512911..f35ef447 100644 --- a/backend/operators/builtin/openvla_finetune/operator.py +++ b/backend/operators/builtin/openvla_finetune/operator.py @@ -2,7 +2,7 @@ from __future__ import annotations from typing import Any -from backend.operators.builtin.command_model_operator import HelperMap, run_command_model_operator +from backend.operators.builtin.structured_model_operator import HelperMap, run_transformers_text_finetune def run( @@ -13,15 +13,4 @@ def run( inputs: list[dict[str, Any]], helpers: HelperMap, ) -> list[dict[str, Any]]: - return run_command_model_operator( - run_id, - node_id, - output_dir, - node_data, - inputs, - helpers, - operator_name="OpenVLA 微调训练", - model_family="OpenVLA LoRA/QLoRA", - recommended_gpu="1x NVIDIA A800 80GB", - recommended_gpu_memory_gb=80, - ) + return run_transformers_text_finetune(run_id, node_id, output_dir, node_data, inputs, helpers) diff --git a/backend/operators/builtin/openvla_finetune/repository.json b/backend/operators/builtin/openvla_finetune/repository.json index 44c72c6b..f8963bdd 100644 --- a/backend/operators/builtin/openvla_finetune/repository.json +++ b/backend/operators/builtin/openvla_finetune/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "AI", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-openvla-finetune:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-openvla-finetune@sha256:36184d088fb2ca79a69c6c11adfba4099a1afa32eac7ec3da6408946044837d2", "metadata": "operator.json", "name": "OpenVLA 微调训练", "operatorId": "operator-builtin-openvla-finetune", @@ -18,6 +25,45 @@ "preferredSlug": "openvla_finetune", "pythonImport": "backend.operators.builtin.openvla_finetune.operator:run", "repositoryId": "builtin/openvla_finetune", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": true, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_openvla_finetune", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-openvla-finetune", + "operatorSlug": "openvla_finetune", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 1, + "recommendedGpuType": "NVIDIA A800 80GB", + "requiresGpu": true + }, + "runtimeDependencies": [ + "openvla-training-command", + "vla-sample-action-manifest", + "gpu-a800" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "openvla-training-command", "vla-sample-action-manifest", diff --git a/backend/operators/builtin/openvla_inference/operator.py b/backend/operators/builtin/openvla_inference/operator.py index 35eceb25..f06cdd58 100644 --- a/backend/operators/builtin/openvla_inference/operator.py +++ b/backend/operators/builtin/openvla_inference/operator.py @@ -2,7 +2,7 @@ from __future__ import annotations from typing import Any -from backend.operators.builtin.command_model_operator import HelperMap, run_command_model_operator +from backend.operators.builtin.structured_model_operator import HelperMap, run_transformers_text_inference def run( @@ -13,15 +13,4 @@ def run( inputs: list[dict[str, Any]], helpers: HelperMap, ) -> list[dict[str, Any]]: - return run_command_model_operator( - run_id, - node_id, - output_dir, - node_data, - inputs, - helpers, - operator_name="OpenVLA 推理", - model_family="OpenVLA inference", - recommended_gpu="1x NVIDIA A800 80GB", - recommended_gpu_memory_gb=40, - ) + return run_transformers_text_inference(run_id, node_id, output_dir, node_data, inputs, helpers) diff --git a/backend/operators/builtin/openvla_inference/repository.json b/backend/operators/builtin/openvla_inference/repository.json index bcefb602..5350b95c 100644 --- a/backend/operators/builtin/openvla_inference/repository.json +++ b/backend/operators/builtin/openvla_inference/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "AI", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-openvla-inference:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-openvla-inference@sha256:3336a89556d87b18a268be0234ff95acd6b23405e90c0da6fe9cdcbf7d3c524e", "metadata": "operator.json", "name": "OpenVLA 推理", "operatorId": "operator-builtin-openvla-inference", @@ -18,6 +25,45 @@ "preferredSlug": "openvla_inference", "pythonImport": "backend.operators.builtin.openvla_inference.operator:run", "repositoryId": "builtin/openvla_inference", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": true, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_openvla_inference", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-openvla-inference", + "operatorSlug": "openvla_inference", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 1, + "recommendedGpuType": "NVIDIA A800 80GB", + "requiresGpu": true + }, + "runtimeDependencies": [ + "openvla-inference-command", + "openvla-model-weights", + "gpu-a800" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "openvla-inference-command", "openvla-model-weights", diff --git a/backend/operators/builtin/package/repository.json b/backend/operators/builtin/package/repository.json index 3869fd3e..92ecc627 100644 --- a/backend/operators/builtin/package/repository.json +++ b/backend/operators/builtin/package/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "VIDEO", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-package:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-package@sha256:e700d9d957364a1446c34f797fbdddd1c9e488c9e866913aeb740cb9b0a03741", "metadata": "operator.json", "name": "成品打包", "operatorId": "operator-builtin-package", @@ -18,6 +25,43 @@ "preferredSlug": "package", "pythonImport": "backend.operators.builtin.package.operator:run", "repositoryId": "builtin/package", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_result_package", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-package", + "operatorSlug": "package", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "filesystem" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "filesystem" ], diff --git a/backend/operators/builtin/parquet_table_summary/repository.json b/backend/operators/builtin/parquet_table_summary/repository.json index 4b29e53e..b6725386 100644 --- a/backend/operators/builtin/parquet_table_summary/repository.json +++ b/backend/operators/builtin/parquet_table_summary/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "DATASET", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-parquet-table-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-parquet-table-summary@sha256:03e9c35aa6e3658bddd4821d6125b5bff10b4d412d33d655dfd869cdae3d03fb", "metadata": "operator.json", "name": "Parquet 表摘要", "operatorId": "operator-builtin-parquet-table-summary", @@ -18,6 +25,44 @@ "preferredSlug": "parquet_table_summary", "pythonImport": "backend.operators.builtin.parquet_table_summary.operator:run", "repositoryId": "builtin/parquet_table_summary", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_parquet_table_summary", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-parquet-table-summary", + "operatorSlug": "parquet_table_summary", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "parquet-files", + "pyarrow" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "parquet-files", "pyarrow" diff --git a/backend/operators/builtin/partnet_mobility_asset_validate/repository.json b/backend/operators/builtin/partnet_mobility_asset_validate/repository.json index 8506480d..13c91162 100644 --- a/backend/operators/builtin/partnet_mobility_asset_validate/repository.json +++ b/backend/operators/builtin/partnet_mobility_asset_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-partnet-mobility-asset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-partnet-mobility-asset-validate@sha256:52bb4ada32074e453f43aaf5fb9f7856a71eeb4e8ae6f2f0751943da55af93a0", "metadata": "operator.json", "name": "PartNet-Mobility 资产校验", "operatorId": "operator-builtin-partnet-mobility-asset-validate", @@ -18,6 +25,47 @@ "preferredSlug": "partnet_mobility_asset_validate", "pythonImport": "backend.operators.builtin.partnet_mobility_asset_validate.operator:run", "repositoryId": "builtin/partnet_mobility_asset_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_partnet_mobility_asset_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-partnet-mobility-asset-validate", + "operatorSlug": "partnet_mobility_asset_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "partnet-mobility-sapien-assets", + "urdf-files", + "json-files", + "semantics-txt", + "mesh-obj-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "partnet-mobility-sapien-assets", "urdf-files", diff --git a/backend/operators/builtin/pointcloud_summary/repository.json b/backend/operators/builtin/pointcloud_summary/repository.json index 453b4307..97b2fc65 100644 --- a/backend/operators/builtin/pointcloud_summary/repository.json +++ b/backend/operators/builtin/pointcloud_summary/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-pointcloud-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-pointcloud-summary@sha256:fa846da15ab5d4816d78375fc700873996fa0ca3a0aaeb9711f106e348146235", "metadata": "operator.json", "name": "点云数据概览", "operatorId": "operator-builtin-pointcloud-summary", @@ -18,6 +25,44 @@ "preferredSlug": "pointcloud_summary", "pythonImport": "backend.operators.builtin.pointcloud_summary.operator:run", "repositoryId": "builtin/pointcloud_summary", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_pointcloud_summary", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-pointcloud-summary", + "operatorSlug": "pointcloud_summary", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "pcd-files", + "ply-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "pcd-files", "ply-files" diff --git a/backend/operators/builtin/real_source/repository.json b/backend/operators/builtin/real_source/repository.json index 7b1f1dc1..4510cf71 100644 --- a/backend/operators/builtin/real_source/repository.json +++ b/backend/operators/builtin/real_source/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "IO", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-real-source:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-real-source@sha256:833d6030b8c10cabfddb3cad54137bb08f232f2bf5071863021276e6b6311d66", "metadata": "operator.json", "name": "真实数据源", "operatorId": "operator-builtin-real-source", @@ -18,6 +25,43 @@ "preferredSlug": "real_source", "pythonImport": "backend.operators.builtin.real_source.operator:run", "repositoryId": "builtin/real_source", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_source", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-real-source", + "operatorSlug": "real_source", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "filesystem" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "filesystem" ], diff --git a/backend/operators/builtin/rgbd_frame_pair_validate/repository.json b/backend/operators/builtin/rgbd_frame_pair_validate/repository.json index 10067df9..c7e72f19 100644 --- a/backend/operators/builtin/rgbd_frame_pair_validate/repository.json +++ b/backend/operators/builtin/rgbd_frame_pair_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rgbd-frame-pair-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rgbd-frame-pair-validate@sha256:4c43bb4d7b81eea7604f3dce8d074201de8187aa2b12cce1e2f7fa1f049838b8", "metadata": "operator.json", "name": "RGB-D 帧对齐校验", "operatorId": "operator-builtin-rgbd-frame-pair-validate", @@ -18,6 +25,47 @@ "preferredSlug": "rgbd_frame_pair_validate", "pythonImport": "backend.operators.builtin.rgbd_frame_pair_validate.operator:run", "repositoryId": "builtin/rgbd_frame_pair_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_rgbd_frame_pair_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-rgbd-frame-pair-validate", + "operatorSlug": "rgbd_frame_pair_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "rgb-images", + "depth-images", + "opencv-python-headless", + "numpy", + "pyyaml" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "rgb-images", "depth-images", diff --git a/backend/operators/builtin/rh20t_scene_validate/repository.json b/backend/operators/builtin/rh20t_scene_validate/repository.json index 7830f1e2..1107b4e5 100644 --- a/backend/operators/builtin/rh20t_scene_validate/repository.json +++ b/backend/operators/builtin/rh20t_scene_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rh20t-scene-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rh20t-scene-validate@sha256:ba1e0190c4d8283480bbed515f6a3ba26132bfc4fb74ec701b137b043f194c94", "metadata": "operator.json", "name": "RH20T 场景结构校验", "operatorId": "operator-builtin-rh20t-scene-validate", @@ -18,6 +25,46 @@ "preferredSlug": "rh20t_scene_validate", "pythonImport": "backend.operators.builtin.rh20t_scene_validate.operator:run", "repositoryId": "builtin/rh20t_scene_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_rh20t_scene_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-rh20t-scene-validate", + "operatorSlug": "rh20t_scene_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "rh20t-scene-directory", + "numpy", + "npy-files", + "image-video-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "rh20t-scene-directory", "numpy", diff --git a/backend/operators/builtin/rlbench_dataset_validate/repository.json b/backend/operators/builtin/rlbench_dataset_validate/repository.json index 054230db..a5e8411e 100644 --- a/backend/operators/builtin/rlbench_dataset_validate/repository.json +++ b/backend/operators/builtin/rlbench_dataset_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rlbench-dataset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rlbench-dataset-validate@sha256:f33e00d9a03bad99b9abeac51c69bf561b766b64188c744a154086364bc75764", "metadata": "operator.json", "name": "RLBench 数据集校验", "operatorId": "operator-builtin-rlbench-dataset-validate", @@ -18,6 +25,45 @@ "preferredSlug": "rlbench_dataset_validate", "pythonImport": "backend.operators.builtin.rlbench_dataset_validate.operator:run", "repositoryId": "builtin/rlbench_dataset_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_rlbench_dataset_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-rlbench-dataset-validate", + "operatorSlug": "rlbench_dataset_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "rlbench-episode-files", + "pickle-files", + "image-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "rlbench-episode-files", "pickle-files", diff --git a/backend/operators/builtin/rlds_episode_summary/repository.json b/backend/operators/builtin/rlds_episode_summary/repository.json index 33dc08d4..675ff61f 100644 --- a/backend/operators/builtin/rlds_episode_summary/repository.json +++ b/backend/operators/builtin/rlds_episode_summary/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rlds-episode-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rlds-episode-summary@sha256:49aae340f8d0d11ebdf89f9a01946477472c9b09b34e7f51fdef51c2feffabcb", "metadata": "operator.json", "name": "RLDS 轨迹概览", "operatorId": "operator-builtin-rlds-episode-summary", @@ -18,6 +25,43 @@ "preferredSlug": "rlds_episode_summary", "pythonImport": "backend.operators.builtin.rlds_episode_summary.operator:run", "repositoryId": "builtin/rlds_episode_summary", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_rlds_episode_summary", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-rlds-episode-summary", + "operatorSlug": "rlds_episode_summary", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "rlds-json-or-tfds-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "rlds-json-or-tfds-files" ], diff --git a/backend/operators/builtin/robocasa_hdf5_demonstration_validate/repository.json b/backend/operators/builtin/robocasa_hdf5_demonstration_validate/repository.json index 3e79c184..d075d789 100644 --- a/backend/operators/builtin/robocasa_hdf5_demonstration_validate/repository.json +++ b/backend/operators/builtin/robocasa_hdf5_demonstration_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-robocasa-hdf5-demonstration-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-robocasa-hdf5-demonstration-validate@sha256:d6010ee3a8123e3b67d406742ac822911ea556f69320ef5603edba24a3b691d5", "metadata": "operator.json", "name": "RoboCasa HDF5示教校验", "operatorId": "operator-builtin-robocasa-hdf5-demonstration-validate", @@ -18,6 +25,45 @@ "preferredSlug": "robocasa_hdf5_demonstration_validate", "pythonImport": "backend.operators.builtin.robocasa_hdf5_demonstration_validate.operator:run", "repositoryId": "builtin/robocasa_hdf5_demonstration_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_robocasa_hdf5_demonstration_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-robocasa-hdf5-demonstration-validate", + "operatorSlug": "robocasa_hdf5_demonstration_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "robocasa-hdf5-demo-files", + "h5py", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "robocasa-hdf5-demo-files", "h5py", diff --git a/backend/operators/builtin/robomimic_hdf5_validate/repository.json b/backend/operators/builtin/robomimic_hdf5_validate/repository.json index 5ad86dea..c450df6e 100644 --- a/backend/operators/builtin/robomimic_hdf5_validate/repository.json +++ b/backend/operators/builtin/robomimic_hdf5_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-robomimic-hdf5-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-robomimic-hdf5-validate@sha256:0d6c7b88a5c3139f3087deb754dbc7642dcbcd2899c1c708a21b1f7c21a04e66", "metadata": "operator.json", "name": "robomimic HDF5 示教校验", "operatorId": "operator-builtin-robomimic-hdf5-validate", @@ -18,6 +25,45 @@ "preferredSlug": "robomimic_hdf5_validate", "pythonImport": "backend.operators.builtin.robomimic_hdf5_validate.operator:run", "repositoryId": "builtin/robomimic_hdf5_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_robomimic_hdf5_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-robomimic-hdf5-validate", + "operatorSlug": "robomimic_hdf5_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "robomimic-hdf5-files", + "h5py", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "robomimic-hdf5-files", "h5py", diff --git a/backend/operators/builtin/robomind_hdf5_quality_validate/repository.json b/backend/operators/builtin/robomind_hdf5_quality_validate/repository.json index d9f91bdd..7156e988 100644 --- a/backend/operators/builtin/robomind_hdf5_quality_validate/repository.json +++ b/backend/operators/builtin/robomind_hdf5_quality_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-robomind-hdf5-quality-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-robomind-hdf5-quality-validate@sha256:e0973f8631bf7234ab5a0073f50f3fb79fc1cf333d0e8060596c218168f2913b", "metadata": "operator.json", "name": "RoboMIND HDF5质量校验", "operatorId": "operator-builtin-robomind-hdf5-quality-validate", @@ -18,6 +25,45 @@ "preferredSlug": "robomind_hdf5_quality_validate", "pythonImport": "backend.operators.builtin.robomind_hdf5_quality_validate.operator:run", "repositoryId": "builtin/robomind_hdf5_quality_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_robomind_hdf5_quality_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-robomind-hdf5-quality-validate", + "operatorSlug": "robomind_hdf5_quality_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "robomind-trajectory-hdf5-files", + "h5py", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "robomind-trajectory-hdf5-files", "h5py", diff --git a/backend/operators/builtin/roboset_hdf5_validate/repository.json b/backend/operators/builtin/roboset_hdf5_validate/repository.json index 5e2abb31..f828a1af 100644 --- a/backend/operators/builtin/roboset_hdf5_validate/repository.json +++ b/backend/operators/builtin/roboset_hdf5_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-roboset-hdf5-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-roboset-hdf5-validate@sha256:db36835317e9ce8c8fa5d550ddf8327460f2a891bf57a2219bd874e6eb154ec1", "metadata": "operator.json", "name": "RoboSet HDF5轨迹校验", "operatorId": "operator-builtin-roboset-hdf5-validate", @@ -18,6 +25,45 @@ "preferredSlug": "roboset_hdf5_validate", "pythonImport": "backend.operators.builtin.roboset_hdf5_validate.operator:run", "repositoryId": "builtin/roboset_hdf5_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_roboset_hdf5_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-roboset-hdf5-validate", + "operatorSlug": "roboset_hdf5_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "roboset-robohive-hdf5-trajectory-files", + "h5py", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "roboset-robohive-hdf5-trajectory-files", "h5py", diff --git a/backend/operators/builtin/robot_model_asset_validate/repository.json b/backend/operators/builtin/robot_model_asset_validate/repository.json index d7c3a4c9..57010fd4 100644 --- a/backend/operators/builtin/robot_model_asset_validate/repository.json +++ b/backend/operators/builtin/robot_model_asset_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-robot-model-asset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-robot-model-asset-validate@sha256:9e85004992d213c08d88dba1c4caf0221e56c6feee9e6e92ab48ce4b4c79fe93", "metadata": "operator.json", "name": "机器人模型资产校验", "operatorId": "operator-builtin-robot-model-asset-validate", @@ -18,6 +25,45 @@ "preferredSlug": "robot_model_asset_validate", "pythonImport": "backend.operators.builtin.robot_model_asset_validate.operator:run", "repositoryId": "builtin/robot_model_asset_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_robot_model_asset_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-robot-model-asset-validate", + "operatorSlug": "robot_model_asset_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "urdf-files", + "mjcf-files", + "mesh-assets" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "urdf-files", "mjcf-files", diff --git a/backend/operators/builtin/roboturk_real_hdf5_validate/repository.json b/backend/operators/builtin/roboturk_real_hdf5_validate/repository.json index b6000827..a8186800 100644 --- a/backend/operators/builtin/roboturk_real_hdf5_validate/repository.json +++ b/backend/operators/builtin/roboturk_real_hdf5_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-roboturk-real-hdf5-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-roboturk-real-hdf5-validate@sha256:db43d615bf52f8ea734d8a45e001602e91b32c7b2046e51cade16489faf4c838", "metadata": "operator.json", "name": "RoboTurk Real HDF5校验", "operatorId": "operator-builtin-roboturk-real-hdf5-validate", @@ -18,6 +25,45 @@ "preferredSlug": "roboturk_real_hdf5_validate", "pythonImport": "backend.operators.builtin.roboturk_real_hdf5_validate.operator:run", "repositoryId": "builtin/roboturk_real_hdf5_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_roboturk_real_hdf5_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-roboturk-real-hdf5-validate", + "operatorSlug": "roboturk_real_hdf5_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "roboturk-real-aligned-hdf5-files", + "h5py", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "roboturk-real-aligned-hdf5-files", "h5py", diff --git a/backend/operators/builtin/ros_joint_state_validate/repository.json b/backend/operators/builtin/ros_joint_state_validate/repository.json index 86039e33..d6e166dd 100644 --- a/backend/operators/builtin/ros_joint_state_validate/repository.json +++ b/backend/operators/builtin/ros_joint_state_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-ros-joint-state-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-ros-joint-state-validate@sha256:55199b510c1c2329facf0a12214c17fb3c4d7c256e89b6772c355a072bd878f7", "metadata": "operator.json", "name": "ROS JointState 校验", "operatorId": "operator-builtin-ros-joint-state-validate", @@ -18,6 +25,45 @@ "preferredSlug": "ros_joint_state_validate", "pythonImport": "backend.operators.builtin.ros_joint_state_validate.operator:run", "repositoryId": "builtin/ros_joint_state_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_ros_joint_state_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-ros-joint-state-validate", + "operatorSlug": "ros_joint_state_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "ros-joint-state-files", + "json-jsonl-csv-parquet-files", + "pyarrow" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "ros-joint-state-files", "json-jsonl-csv-parquet-files", diff --git a/backend/operators/builtin/rosbag_summary/repository.json b/backend/operators/builtin/rosbag_summary/repository.json index 9041ea65..4dadc8d8 100644 --- a/backend/operators/builtin/rosbag_summary/repository.json +++ b/backend/operators/builtin/rosbag_summary/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rosbag-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rosbag-summary@sha256:9d41ed0073c84068869b7c844cf2a5362601dab551835a0ce0a3b1a9d00de9c5", "metadata": "operator.json", "name": "ROSBag 数据概览", "operatorId": "operator-builtin-rosbag-summary", @@ -18,6 +25,44 @@ "preferredSlug": "rosbag_summary", "pythonImport": "backend.operators.builtin.rosbag_summary.operator:run", "repositoryId": "builtin/rosbag_summary", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_rosbag_summary", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-rosbag-summary", + "operatorSlug": "rosbag_summary", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "rosbag2-files", + "sqlite" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "rosbag2-files", "sqlite" diff --git a/backend/operators/builtin/rtx_action_schema_validate/repository.json b/backend/operators/builtin/rtx_action_schema_validate/repository.json index dbe3780c..b1ea5c94 100644 --- a/backend/operators/builtin/rtx_action_schema_validate/repository.json +++ b/backend/operators/builtin/rtx_action_schema_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rtx-action-schema-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rtx-action-schema-validate@sha256:65dc02ffb329e3df1a0cadf66dc92379c92b3e98b6d35bfe460f6ea893aea031", "metadata": "operator.json", "name": "RT-X动作空间Schema校验", "operatorId": "operator-builtin-rtx-action-schema-validate", @@ -18,6 +25,46 @@ "preferredSlug": "rtx_action_schema_validate", "pythonImport": "backend.operators.builtin.rtx_action_schema_validate.operator:run", "repositoryId": "builtin/rtx_action_schema_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_rtx_action_schema_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-rtx-action-schema-validate", + "operatorSlug": "rtx_action_schema_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "csv-json-jsonl-parquet-hdf5-action-files", + "pyarrow", + "h5py", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "csv-json-jsonl-parquet-hdf5-action-files", "pyarrow", diff --git a/backend/operators/builtin/segmentation_mask_validate/repository.json b/backend/operators/builtin/segmentation_mask_validate/repository.json index 29950f7e..02ea0476 100644 --- a/backend/operators/builtin/segmentation_mask_validate/repository.json +++ b/backend/operators/builtin/segmentation_mask_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-segmentation-mask-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-segmentation-mask-validate@sha256:c96e3c47fe7bab282925c75d473fd1e8c400605c5d33b9dac9ab68ca98e4fbd3", "metadata": "operator.json", "name": "分割 Mask 校验", "operatorId": "operator-builtin-segmentation-mask-validate", @@ -18,6 +25,45 @@ "preferredSlug": "segmentation_mask_validate", "pythonImport": "backend.operators.builtin.segmentation_mask_validate.operator:run", "repositoryId": "builtin/segmentation_mask_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_segmentation_mask_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-segmentation-mask-validate", + "operatorSlug": "segmentation_mask_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "segmentation-mask-images", + "opencv-python-headless", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "segmentation-mask-images", "opencv-python-headless", diff --git a/backend/operators/builtin/sensor_timestamp_sync_validate/repository.json b/backend/operators/builtin/sensor_timestamp_sync_validate/repository.json index 5d78fda1..ee42917e 100644 --- a/backend/operators/builtin/sensor_timestamp_sync_validate/repository.json +++ b/backend/operators/builtin/sensor_timestamp_sync_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-sensor-timestamp-sync-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-sensor-timestamp-sync-validate@sha256:4afd1efaf170d73917f215fbfdef55f36869726a32a07e79b9037204febe9a8b", "metadata": "operator.json", "name": "多传感器时间同步校验", "operatorId": "operator-builtin-sensor-timestamp-sync-validate", @@ -18,6 +25,43 @@ "preferredSlug": "sensor_timestamp_sync_validate", "pythonImport": "backend.operators.builtin.sensor_timestamp_sync_validate.operator:run", "repositoryId": "builtin/sensor_timestamp_sync_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_sensor_timestamp_sync_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-sensor-timestamp-sync-validate", + "operatorSlug": "sensor_timestamp_sync_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "timestamp-csv-jsonl" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "timestamp-csv-jsonl" ], diff --git a/backend/operators/builtin/shell_command/repository.json b/backend/operators/builtin/shell_command/repository.json index 1809e064..60daa1dc 100644 --- a/backend/operators/builtin/shell_command/repository.json +++ b/backend/operators/builtin/shell_command/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "UTIL", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-shell-command:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-shell-command@sha256:47f7039f4ddcf58fb51cf38f1e4fe8fd7c96e8bcad2ea745c88ee0d9e1423f51", "metadata": "operator.json", "name": "Shell 命令执行", "operatorId": "operator-builtin-shell-command", @@ -18,6 +25,43 @@ "preferredSlug": "shell_command", "pythonImport": "backend.operators.builtin.shell_command.operator:run", "repositoryId": "builtin/shell_command", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_shell_command", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-shell-command", + "operatorSlug": "shell_command", + "parameters": { + "commandWrapperAllowed": true, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "container-shell" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "container-shell" ], diff --git a/backend/operators/builtin/sim3d_asset_generate/repository.json b/backend/operators/builtin/sim3d_asset_generate/repository.json index 02fe00a2..2cd7dfe4 100644 --- a/backend/operators/builtin/sim3d_asset_generate/repository.json +++ b/backend/operators/builtin/sim3d_asset_generate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "SIMULATION", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-sim3d-asset-generate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-sim3d-asset-generate@sha256:c56c68886041a19492d9427211dc4741574a95eacc3242d4f34c1adb6d2bcc36", "metadata": "operator.json", "name": "3D 资产仿真生成", "operatorId": "operator-builtin-sim3d-asset-generate", @@ -18,6 +25,44 @@ "preferredSlug": "sim3d_asset_generate", "pythonImport": "backend.operators.builtin.sim3d_asset_generate.operator:run", "repositoryId": "builtin/sim3d_asset_generate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_sim3d_asset_generate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-sim3d-asset-generate", + "operatorSlug": "sim3d_asset_generate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-artifacts", + "obj-mtl-gltf-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "json-artifacts", "obj-mtl-gltf-files" diff --git a/backend/operators/builtin/structured_model_operator.py b/backend/operators/builtin/structured_model_operator.py new file mode 100644 index 00000000..a2d76dbb --- /dev/null +++ b/backend/operators/builtin/structured_model_operator.py @@ -0,0 +1,383 @@ +from __future__ import annotations + +import csv +import json +import math +import pathlib +import shutil +from typing import Any, Callable + + +HelperMap = dict[str, Callable[..., Any]] + + +def _text(params: dict[str, str], key: str, default: str = "") -> str: + return str(params.get(key, default) or default).strip() + + +def _int(params: dict[str, str], key: str, default: int) -> int: + value = _text(params, key, str(default)) + try: + parsed = int(value) + except ValueError as exc: + raise RuntimeError(f"{key} must be an integer") from exc + if parsed < 1: + raise RuntimeError(f"{key} must be >= 1") + return parsed + + +def _float(params: dict[str, str], key: str, default: float) -> float: + value = _text(params, key, str(default)) + try: + parsed = float(value) + except ValueError as exc: + raise RuntimeError(f"{key} must be a number") from exc + if not math.isfinite(parsed): + raise RuntimeError(f"{key} must be finite") + return parsed + + +def _dataset_path(params: dict[str, str], inputs: list[dict[str, Any]]) -> pathlib.Path: + explicit = _text(params, "DATASET_PATH") or _text(params, "TRAIN_DATA_PATH") or _text(params, "SOURCE_PATH") + if explicit: + path = pathlib.Path(explicit) + if not path.exists(): + raise RuntimeError(f"dataset path does not exist: {path}") + return path + for artifact in inputs: + path = pathlib.Path(str(artifact.get("path") or artifact.get("uri") or "")) + if path.exists(): + return path + raise RuntimeError("structured model operator requires DATASET_PATH/TRAIN_DATA_PATH/SOURCE_PATH or an input artifact path") + + +def _load_json_records(path: pathlib.Path) -> list[dict[str, Any]]: + if path.is_dir(): + candidates = [ + path / "samples.jsonl", + path / "samples.json", + path / "dataset.jsonl", + path / "dataset.json", + path / "manifest.jsonl", + path / "manifest.json", + ] + for candidate in candidates: + if candidate.exists(): + return _load_json_records(candidate) + raise RuntimeError(f"dataset directory does not contain a supported JSON/JSONL manifest: {path}") + if path.suffix.lower() == ".jsonl": + records: list[dict[str, Any]] = [] + for line in path.read_text(encoding="utf-8").splitlines(): + if line.strip(): + payload = json.loads(line) + if isinstance(payload, dict): + records.append(payload) + return records + if path.suffix.lower() == ".csv": + with path.open(newline="", encoding="utf-8") as handle: + return [dict(row) for row in csv.DictReader(handle)] + payload = json.loads(path.read_text(encoding="utf-8")) + if isinstance(payload, dict): + for key in ("samples", "items", "records", "data"): + value = payload.get(key) + if isinstance(value, list): + return [item for item in value if isinstance(item, dict)] + return [payload] + if isinstance(payload, list): + return [item for item in payload if isinstance(item, dict)] + raise RuntimeError(f"dataset must be a JSON object/list, JSONL, or CSV: {path}") + + +def _vector(value: Any) -> list[float]: + if isinstance(value, str): + try: + value = json.loads(value) + except json.JSONDecodeError: + value = [float(part) for part in value.replace(";", ",").split(",") if part.strip()] + if isinstance(value, (int, float)): + value = [value] + if not isinstance(value, list) or not value: + raise RuntimeError("state/action fields must be numeric arrays") + vector = [float(item) for item in value] + if not all(math.isfinite(item) for item in vector): + raise RuntimeError("state/action vectors must contain finite values") + return vector + + +def _field_value(item: dict[str, Any], field: str) -> Any: + value: Any = item + for part in field.split("."): + if isinstance(value, dict): + value = value.get(part) + else: + return None + return value + + +def run_policy_training( + run_id: str, + node_id: str, + output_dir: Any, + node_data: dict[str, Any], + inputs: list[dict[str, Any]], + helpers: HelperMap, + *, + operator_name: str, + model_family: str, +) -> list[dict[str, Any]]: + try: + import torch + from torch import nn + from torch.utils.data import DataLoader, TensorDataset + except Exception as exc: # pragma: no cover - exercised in container images + raise RuntimeError("PyTorch is required in the operator image for structured policy training") from exc + + params = helpers["hyperparams"](node_data) + dataset_path = _dataset_path(params, inputs) + records = _load_json_records(dataset_path) + if len(records) < 2: + raise RuntimeError("policy training requires at least 2 real samples") + state_field = _text(params, "STATE_FIELD", "state") + action_field = _text(params, "ACTION_FIELD", "action") + states = [_vector(_field_value(item, state_field)) for item in records] + actions = [_vector(_field_value(item, action_field)) for item in records] + state_dim = len(states[0]) + action_dim = len(actions[0]) + if any(len(item) != state_dim for item in states) or any(len(item) != action_dim for item in actions): + raise RuntimeError("all state/action vectors must have consistent dimensions") + + epochs = _int(params, "EPOCHS", 5) + batch_size = _int(params, "BATCH_SIZE", 8) + learning_rate = _float(params, "LEARNING_RATE", 0.001) + hidden_dim = _int(params, "HIDDEN_DIM", 128) + chunk_size = _int(params, "CHUNK_SIZE", 1) + + x = torch.tensor(states, dtype=torch.float32) + y = torch.tensor(actions, dtype=torch.float32) + loader = DataLoader(TensorDataset(x, y), batch_size=min(batch_size, len(records)), shuffle=True) + model = nn.Sequential(nn.Linear(state_dim, hidden_dim), nn.ReLU(), nn.Linear(hidden_dim, action_dim)) + optimizer = torch.optim.AdamW(model.parameters(), lr=learning_rate) + loss_fn = nn.MSELoss() + losses: list[float] = [] + for _ in range(epochs): + epoch_loss = 0.0 + for batch_x, batch_y in loader: + optimizer.zero_grad(set_to_none=True) + loss = loss_fn(model(batch_x), batch_y) + loss.backward() + optimizer.step() + epoch_loss += float(loss.detach().cpu()) + losses.append(epoch_loss / max(len(loader), 1)) + + output_path = pathlib.Path(output_dir) + checkpoint_path = output_path / "checkpoint.pt" + torch.save( + { + "operator": operator_name, + "modelFamily": model_family, + "stateField": state_field, + "actionField": action_field, + "stateDim": state_dim, + "actionDim": action_dim, + "chunkSize": chunk_size, + "modelStateDict": model.state_dict(), + }, + checkpoint_path, + ) + metrics = { + "operator": operator_name, + "modelFamily": model_family, + "datasetPath": str(dataset_path), + "sampleCount": len(records), + "stateDim": state_dim, + "actionDim": action_dim, + "epochs": epochs, + "batchSize": batch_size, + "learningRate": learning_rate, + "chunkSize": chunk_size, + "finalLoss": losses[-1], + "losses": losses, + } + return [ + helpers["write_json_artifact"](run_id, node_id, output_path, "metrics.json", metrics), + helpers["build_artifact"](run_id, node_id, checkpoint_path, "model"), + ] + + +def run_yolo_training( + run_id: str, + node_id: str, + output_dir: Any, + node_data: dict[str, Any], + inputs: list[dict[str, Any]], + helpers: HelperMap, +) -> list[dict[str, Any]]: + try: + from ultralytics import YOLO + except Exception as exc: # pragma: no cover - exercised in container images + raise RuntimeError("Ultralytics is required in the YOLO operator image") from exc + + params = helpers["hyperparams"](node_data) + data_yaml = pathlib.Path(_text(params, "DATA_YAML") or _text(params, "DATASET_YAML") or str(_dataset_path(params, inputs))) + if not data_yaml.exists(): + raise RuntimeError(f"YOLO data.yaml does not exist: {data_yaml}") + model_name = _text(params, "MODEL_SIZE", "yolo11n.pt") + epochs = _int(params, "EPOCHS", 10) + image_size = _int(params, "IMAGE_SIZE", 640) + batch_size = _int(params, "BATCH_SIZE", 8) + project_dir = pathlib.Path(output_dir) / "ultralytics" + result = YOLO(model_name).train( + data=str(data_yaml), + epochs=epochs, + imgsz=image_size, + batch=batch_size, + project=str(project_dir), + name="train", + exist_ok=True, + ) + weights_dir = project_dir / "train" / "weights" + best = weights_dir / "best.pt" + last = weights_dir / "last.pt" + checkpoint = best if best.exists() else last + if not checkpoint.exists(): + raise RuntimeError("YOLO training did not produce a checkpoint") + copied_checkpoint = pathlib.Path(output_dir) / checkpoint.name + shutil.copy2(checkpoint, copied_checkpoint) + metrics = { + "operator": "YOLO 具身目标检测训练", + "modelFamily": "YOLO detection", + "dataYaml": str(data_yaml), + "model": model_name, + "epochs": epochs, + "imageSize": image_size, + "batchSize": batch_size, + "checkpoint": str(copied_checkpoint), + "results": str(getattr(result, "save_dir", project_dir / "train")), + } + return [ + helpers["write_json_artifact"](run_id, node_id, pathlib.Path(output_dir), "metrics.json", metrics), + helpers["build_artifact"](run_id, node_id, copied_checkpoint, "model"), + ] + + +def run_transformers_text_finetune( + run_id: str, + node_id: str, + output_dir: Any, + node_data: dict[str, Any], + inputs: list[dict[str, Any]], + helpers: HelperMap, +) -> list[dict[str, Any]]: + try: + import torch + from torch.utils.data import DataLoader, Dataset + from transformers import AutoModelForCausalLM, AutoTokenizer + except Exception as exc: # pragma: no cover - exercised in container images + raise RuntimeError("Transformers and PyTorch are required in the OpenVLA fine-tune operator image") from exc + + params = helpers["hyperparams"](node_data) + base_model = _text(params, "BASE_MODEL") + if not base_model: + raise RuntimeError("BASE_MODEL is required for OpenVLA fine-tune") + dataset_path = _dataset_path(params, inputs) + records = _load_json_records(dataset_path) + text_field = _text(params, "TEXT_FIELD", "instruction") + target_field = _text(params, "TARGET_FIELD", "action") + max_length = _int(params, "MAX_LENGTH", 256) + epochs = _int(params, "EPOCHS", 1) + learning_rate = _float(params, "LEARNING_RATE", 0.00005) + samples = [ + f"{item.get(text_field, '')}\n{item.get(target_field, '')}".strip() + for item in records + if str(item.get(text_field, "")).strip() and str(item.get(target_field, "")).strip() + ] + if not samples: + raise RuntimeError("OpenVLA fine-tune dataset must contain text/action samples") + + tokenizer = AutoTokenizer.from_pretrained(base_model) + if tokenizer.pad_token is None: + tokenizer.pad_token = tokenizer.eos_token + model = AutoModelForCausalLM.from_pretrained(base_model) + + class TextDataset(Dataset): + def __len__(self) -> int: + return len(samples) + + def __getitem__(self, index: int) -> dict[str, Any]: + encoded = tokenizer(samples[index], truncation=True, padding="max_length", max_length=max_length, return_tensors="pt") + item = {key: value.squeeze(0) for key, value in encoded.items()} + item["labels"] = item["input_ids"].clone() + return item + + loader = DataLoader(TextDataset(), batch_size=min(_int(params, "BATCH_SIZE", 1), len(samples)), shuffle=True) + optimizer = torch.optim.AdamW(model.parameters(), lr=learning_rate) + model.train() + losses: list[float] = [] + for _ in range(epochs): + epoch_loss = 0.0 + for batch in loader: + optimizer.zero_grad(set_to_none=True) + output = model(**batch) + output.loss.backward() + optimizer.step() + epoch_loss += float(output.loss.detach().cpu()) + losses.append(epoch_loss / max(len(loader), 1)) + + model_dir = pathlib.Path(output_dir) / "openvla-adapter" + model.save_pretrained(model_dir) + tokenizer.save_pretrained(model_dir) + metrics = { + "operator": "OpenVLA 微调训练", + "modelFamily": "OpenVLA CausalLM fine-tune", + "baseModel": base_model, + "datasetPath": str(dataset_path), + "sampleCount": len(samples), + "epochs": epochs, + "finalLoss": losses[-1], + "losses": losses, + "modelDir": str(model_dir), + } + return [ + helpers["write_json_artifact"](run_id, node_id, pathlib.Path(output_dir), "metrics.json", metrics), + helpers["build_artifact"](run_id, node_id, model_dir, "model"), + ] + + +def run_transformers_text_inference( + run_id: str, + node_id: str, + output_dir: Any, + node_data: dict[str, Any], + inputs: list[dict[str, Any]], + helpers: HelperMap, +) -> list[dict[str, Any]]: + try: + from transformers import AutoModelForCausalLM, AutoTokenizer + except Exception as exc: # pragma: no cover - exercised in container images + raise RuntimeError("Transformers is required in the OpenVLA inference operator image") from exc + + params = helpers["hyperparams"](node_data) + model_path = _text(params, "MODEL_PATH") or _text(params, "BASE_MODEL") + if not model_path: + raise RuntimeError("MODEL_PATH or BASE_MODEL is required for OpenVLA inference") + prompt = _text(params, "PROMPT") + if not prompt: + records = _load_json_records(_dataset_path(params, inputs)) + prompt = str(records[0].get(_text(params, "TEXT_FIELD", "instruction"), "")).strip() if records else "" + if not prompt: + raise RuntimeError("OpenVLA inference requires PROMPT or an input sample instruction") + max_new_tokens = _int(params, "MAX_NEW_TOKENS", 64) + tokenizer = AutoTokenizer.from_pretrained(model_path) + model = AutoModelForCausalLM.from_pretrained(model_path) + encoded = tokenizer(prompt, return_tensors="pt") + generated = model.generate(**encoded, max_new_tokens=max_new_tokens) + text = tokenizer.decode(generated[0], skip_special_tokens=True) + payload = { + "operator": "OpenVLA 推理", + "modelFamily": "OpenVLA CausalLM inference", + "modelPath": model_path, + "prompt": prompt, + "maxNewTokens": max_new_tokens, + "prediction": text, + } + return [helpers["write_json_artifact"](run_id, node_id, pathlib.Path(output_dir), "inference-result.json", payload)] diff --git a/backend/operators/builtin/template_render/repository.json b/backend/operators/builtin/template_render/repository.json index d45f95e8..331bc5b9 100644 --- a/backend/operators/builtin/template_render/repository.json +++ b/backend/operators/builtin/template_render/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "UTIL", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-template-render:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-template-render@sha256:b96fcf95d3500eb6849cdb61df7a5ae6f4d3266fc5e69d6b9cce6ce2a05c8fc4", "metadata": "operator.json", "name": "文本模板渲染", "operatorId": "operator-builtin-template-render", @@ -18,6 +25,43 @@ "preferredSlug": "template_render", "pythonImport": "backend.operators.builtin.template_render.operator:run", "repositoryId": "builtin/template_render", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_text_template", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-template-render", + "operatorSlug": "template_render", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-artifacts" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "json-artifacts" ], diff --git a/backend/operators/builtin/tf_transform_tree_validate/repository.json b/backend/operators/builtin/tf_transform_tree_validate/repository.json index 68b311a3..8b7aaf34 100644 --- a/backend/operators/builtin/tf_transform_tree_validate/repository.json +++ b/backend/operators/builtin/tf_transform_tree_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-tf-transform-tree-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-tf-transform-tree-validate@sha256:4777877db8114d07f770211c4d9b3375300bcbf52823d0abd5696fd2747bb4bc", "metadata": "operator.json", "name": "TF 坐标树校验", "operatorId": "operator-builtin-tf-transform-tree-validate", @@ -18,6 +25,44 @@ "preferredSlug": "tf_transform_tree_validate", "pythonImport": "backend.operators.builtin.tf_transform_tree_validate.operator:run", "repositoryId": "builtin/tf_transform_tree_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_tf_transform_tree_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-tf-transform-tree-validate", + "operatorSlug": "tf_transform_tree_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-jsonl-csv-yaml-files", + "optional-pyyaml" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "json-jsonl-csv-yaml-files", "optional-pyyaml" diff --git a/backend/operators/builtin/tfrecord_structural_validate/repository.json b/backend/operators/builtin/tfrecord_structural_validate/repository.json index 79fda8df..7a5e8d2f 100644 --- a/backend/operators/builtin/tfrecord_structural_validate/repository.json +++ b/backend/operators/builtin/tfrecord_structural_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-tfrecord-structural-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-tfrecord-structural-validate@sha256:60fb1eb05a35a6b7712c5e07ce067962ed125ef3f3e8b73853c56c5d250e30ad", "metadata": "operator.json", "name": "TFRecord 结构校验", "operatorId": "operator-builtin-tfrecord-structural-validate", @@ -18,6 +25,44 @@ "preferredSlug": "tfrecord_structural_validate", "pythonImport": "backend.operators.builtin.tfrecord_structural_validate.operator:run", "repositoryId": "builtin/tfrecord_structural_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_tfrecord_structural_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-tfrecord-structural-validate", + "operatorSlug": "tfrecord_structural_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "tfrecord-files", + "crc32c-structural-validation" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "tfrecord-files", "crc32c-structural-validation" diff --git a/backend/operators/builtin/trajectory_integrity_validate/repository.json b/backend/operators/builtin/trajectory_integrity_validate/repository.json index da345734..79f8cfae 100644 --- a/backend/operators/builtin/trajectory_integrity_validate/repository.json +++ b/backend/operators/builtin/trajectory_integrity_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-trajectory-integrity-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-trajectory-integrity-validate@sha256:1ec455b21a6a4cf3c35953d50d8965c77deeffd71412ffa1ebb3bc66f42fbe9e", "metadata": "operator.json", "name": "轨迹多模态完整性校验", "operatorId": "operator-builtin-trajectory-integrity-validate", @@ -18,6 +25,44 @@ "preferredSlug": "trajectory_integrity_validate", "pythonImport": "backend.operators.builtin.trajectory_integrity_validate.operator:run", "repositoryId": "builtin/trajectory_integrity_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_trajectory_integrity_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-trajectory-integrity-validate", + "operatorSlug": "trajectory_integrity_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "trajectory-json", + "local-media-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "trajectory-json", "local-media-files" diff --git a/backend/operators/builtin/umi_zarr_dataset_validate/repository.json b/backend/operators/builtin/umi_zarr_dataset_validate/repository.json index fbba5d7e..8361d798 100644 --- a/backend/operators/builtin/umi_zarr_dataset_validate/repository.json +++ b/backend/operators/builtin/umi_zarr_dataset_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-umi-zarr-dataset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-umi-zarr-dataset-validate@sha256:a6f4a48ad0f019a3720905ccbc5b8ab733baf6cf03cb0163695205d792831ff1", "metadata": "operator.json", "name": "UMI Zarr数据集校验", "operatorId": "operator-builtin-umi-zarr-dataset-validate", @@ -18,6 +25,45 @@ "preferredSlug": "umi_zarr_dataset_validate", "pythonImport": "backend.operators.builtin.umi_zarr_dataset_validate.operator:run", "repositoryId": "builtin/umi_zarr_dataset_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_umi_zarr_dataset_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-umi-zarr-dataset-validate", + "operatorSlug": "umi_zarr_dataset_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "umi-zarr-dataset", + "zarr", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "umi-zarr-dataset", "zarr", diff --git a/backend/operators/builtin/upload/repository.json b/backend/operators/builtin/upload/repository.json index 75790f11..b73b70de 100644 --- a/backend/operators/builtin/upload/repository.json +++ b/backend/operators/builtin/upload/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "IO", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-upload:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-upload@sha256:bc56ba702686c5ed78cc2ab77ede4fa2476acba81df1aa1bd3ea5d6c7ef88474", "metadata": "operator.json", "name": "结果上传", "operatorId": "operator-builtin-upload", @@ -18,6 +25,44 @@ "preferredSlug": "upload", "pythonImport": "backend.operators.builtin.upload.operator:run", "repositoryId": "builtin/upload", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_upload", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-upload", + "operatorSlug": "upload", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "filesystem", + "optional-baidu-bos" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "filesystem", "optional-baidu-bos" diff --git a/backend/operators/builtin/video_annotate/repository.json b/backend/operators/builtin/video_annotate/repository.json index c950bacd..4d313378 100644 --- a/backend/operators/builtin/video_annotate/repository.json +++ b/backend/operators/builtin/video_annotate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "VIDEO", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-annotate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-annotate@sha256:b2bb1476b0994f852a057c1b81bc412ed25c257f3581b720752c19435045f06d", "metadata": "operator.json", "name": "视频标注", "operatorId": "operator-builtin-video-annotate", @@ -18,6 +25,44 @@ "preferredSlug": "video_annotate", "pythonImport": "backend.operators.builtin.video_annotate.operator:run", "repositoryId": "builtin/video_annotate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_video_annotation", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-video-annotate", + "operatorSlug": "video_annotate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "video-files", + "json-artifacts" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "video-files", "json-artifacts" diff --git a/backend/operators/builtin/video_filter/repository.json b/backend/operators/builtin/video_filter/repository.json index 2eb734bd..acad69f0 100644 --- a/backend/operators/builtin/video_filter/repository.json +++ b/backend/operators/builtin/video_filter/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "VIDEO", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-filter:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-filter@sha256:421e6b00389e9761b1bec6b316dec38151200f055ff9bd6deb6659ce170f97ae", "metadata": "operator.json", "name": "数据过滤", "operatorId": "operator-builtin-video-filter", @@ -18,6 +25,43 @@ "preferredSlug": "video_filter", "pythonImport": "backend.operators.builtin.video_filter.operator:run", "repositoryId": "builtin/video_filter", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_dataset_filter", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-video-filter", + "operatorSlug": "video_filter", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-artifacts" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "json-artifacts" ], diff --git a/backend/operators/builtin/video_quality/repository.json b/backend/operators/builtin/video_quality/repository.json index 2a7977ae..537c88d0 100644 --- a/backend/operators/builtin/video_quality/repository.json +++ b/backend/operators/builtin/video_quality/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "VIDEO", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-quality:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-quality@sha256:755209a0ec2c869a9784e70fa4cad0803dda72f031791f1e43363bc38b3c4ece", "metadata": "operator.json", "name": "视频质检", "operatorId": "operator-builtin-video-quality", @@ -18,6 +25,44 @@ "preferredSlug": "video_quality", "pythonImport": "backend.operators.builtin.video_quality.operator:run", "repositoryId": "builtin/video_quality", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_video_quality", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-video-quality", + "operatorSlug": "video_quality", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "ffprobe", + "video-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "ffprobe", "video-files" diff --git a/backend/operators/builtin/video_slice/repository.json b/backend/operators/builtin/video_slice/repository.json index 76e153b9..2f4d2ca5 100644 --- a/backend/operators/builtin/video_slice/repository.json +++ b/backend/operators/builtin/video_slice/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "VIDEO", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-slice:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-slice@sha256:0102dbe48263fd4e6a62ab9ed48619d10a91c9b4f201ad0dc5b06071c62b05c1", "metadata": "operator.json", "name": "视频切片", "operatorId": "operator-builtin-video-slice", @@ -18,6 +25,44 @@ "preferredSlug": "video_slice", "pythonImport": "backend.operators.builtin.video_slice.operator:run", "repositoryId": "builtin/video_slice", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_video_slicing", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-video-slice", + "operatorSlug": "video_slice", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "ffmpeg", + "video-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "ffmpeg", "video-files" diff --git a/backend/operators/builtin/video_target_clip/repository.json b/backend/operators/builtin/video_target_clip/repository.json index 8be53ca8..c0aa4eef 100644 --- a/backend/operators/builtin/video_target_clip/repository.json +++ b/backend/operators/builtin/video_target_clip/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "VIDEO", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-target-clip:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-target-clip@sha256:0eaf4f0fbd20a6f017f1dd77a7d4660dc5cac1112dbbb32fc8e1a4c0cf6377f1", "metadata": "operator.json", "name": "视频目标检测裁剪", "operatorId": "operator-builtin-video-target-clip", @@ -18,6 +25,47 @@ "preferredSlug": "video_target_clip", "pythonImport": "backend.operators.builtin.video_target_clip.operator:run", "repositoryId": "builtin/video_target_clip", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_video_target_clip", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-video-target-clip", + "operatorSlug": "video_target_clip", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "opencv-python-headless", + "numpy", + "ffmpeg", + "video-files", + "optional-opencv-dnn-model" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "opencv-python-headless", "numpy", diff --git a/backend/operators/builtin/vla_model_evaluate/repository.json b/backend/operators/builtin/vla_model_evaluate/repository.json index d0a0ad4a..21a676c8 100644 --- a/backend/operators/builtin/vla_model_evaluate/repository.json +++ b/backend/operators/builtin/vla_model_evaluate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "AI", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-vla-model-evaluate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-vla-model-evaluate@sha256:d940646dabe0e5b7b950e1f32a0ce36110d9dbc379bd123dbb761b8f2e896464", "metadata": "operator.json", "name": "轻量策略评测", "operatorId": "operator-builtin-vla-model-evaluate", @@ -18,6 +25,44 @@ "preferredSlug": "vla_model_evaluate", "pythonImport": "backend.operators.builtin.vla_model_evaluate.operator:run", "repositoryId": "builtin/vla_model_evaluate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_vla_model_evaluate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-vla-model-evaluate", + "operatorSlug": "vla_model_evaluate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "vla-inference-result-json", + "vla-sample-action-manifest" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "vla-inference-result-json", "vla-sample-action-manifest" diff --git a/backend/operators/builtin/vla_model_inference/repository.json b/backend/operators/builtin/vla_model_inference/repository.json index a7f4cc58..870c24d9 100644 --- a/backend/operators/builtin/vla_model_inference/repository.json +++ b/backend/operators/builtin/vla_model_inference/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "AI", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-vla-model-inference:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-vla-model-inference@sha256:6e63bbeca4792ae37589211f3244c0545cbe2217e6f6bcf709b0575275d7d209", "metadata": "operator.json", "name": "轻量线性策略推理", "operatorId": "operator-builtin-vla-model-inference", @@ -18,6 +25,44 @@ "preferredSlug": "vla_model_inference", "pythonImport": "backend.operators.builtin.vla_model_inference.operator:run", "repositoryId": "builtin/vla_model_inference", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_vla_model_inference", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-vla-model-inference", + "operatorSlug": "vla_model_inference", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "vla-policy-model-json", + "vla-sample-action-manifest" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "vla-policy-model-json", "vla-sample-action-manifest" diff --git a/backend/operators/builtin/vla_model_train/repository.json b/backend/operators/builtin/vla_model_train/repository.json index 6aa28c2c..2f76fe70 100644 --- a/backend/operators/builtin/vla_model_train/repository.json +++ b/backend/operators/builtin/vla_model_train/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "AI", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-vla-model-train:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-vla-model-train@sha256:eccc4d24b9a8f0a69d9de173f3eb7b4541550245138babdd68571a77cd91fc45", "metadata": "operator.json", "name": "轻量线性策略训练", "operatorId": "operator-builtin-vla-model-train", @@ -18,6 +25,44 @@ "preferredSlug": "vla_model_train", "pythonImport": "backend.operators.builtin.vla_model_train.operator:run", "repositoryId": "builtin/vla_model_train", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_vla_model_train", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-vla-model-train", + "operatorSlug": "vla_model_train", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "vla-sample-action-manifest", + "json-jsonl-csv-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "vla-sample-action-manifest", "json-jsonl-csv-files" diff --git a/backend/operators/builtin/vla_sample_manifest_validate/repository.json b/backend/operators/builtin/vla_sample_manifest_validate/repository.json index f9c575e1..88843373 100644 --- a/backend/operators/builtin/vla_sample_manifest_validate/repository.json +++ b/backend/operators/builtin/vla_sample_manifest_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-vla-sample-manifest-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-vla-sample-manifest-validate@sha256:bcfb95b9a40b9152f2d102c9a00b18e157281e6d7438c84cd7ec6dfb6160df69", "metadata": "operator.json", "name": "VLA 样本清单校验", "operatorId": "operator-builtin-vla-sample-manifest-validate", @@ -18,6 +25,45 @@ "preferredSlug": "vla_sample_manifest_validate", "pythonImport": "backend.operators.builtin.vla_sample_manifest_validate.operator:run", "repositoryId": "builtin/vla_sample_manifest_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_vla_sample_manifest_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-vla-sample-manifest-validate", + "operatorSlug": "vla_sample_manifest_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-jsonl-csv-parquet-files", + "image-files", + "pyarrow" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "json-jsonl-csv-parquet-files", "image-files", diff --git a/backend/operators/builtin/webdataset_shard_summary/repository.json b/backend/operators/builtin/webdataset_shard_summary/repository.json index 5387658c..b8f102dd 100644 --- a/backend/operators/builtin/webdataset_shard_summary/repository.json +++ b/backend/operators/builtin/webdataset_shard_summary/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "DATASET", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-webdataset-shard-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-webdataset-shard-summary@sha256:ce71042733709b064da7e6ceeadfa4a3f4c138d627da5ff9fe7b877d1de2b375", "metadata": "operator.json", "name": "WebDataset 分片概览", "operatorId": "operator-builtin-webdataset-shard-summary", @@ -18,6 +25,44 @@ "preferredSlug": "webdataset_shard_summary", "pythonImport": "backend.operators.builtin.webdataset_shard_summary.operator:run", "repositoryId": "builtin/webdataset_shard_summary", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_webdataset_shard_summary", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-webdataset-shard-summary", + "operatorSlug": "webdataset_shard_summary", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "webdataset-tar-shards", + "python-tarfile" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "webdataset-tar-shards", "python-tarfile" diff --git a/backend/operators/builtin/xplatform_dataset_index/repository.json b/backend/operators/builtin/xplatform_dataset_index/repository.json index 93b19d99..0bed5b66 100644 --- a/backend/operators/builtin/xplatform_dataset_index/repository.json +++ b/backend/operators/builtin/xplatform_dataset_index/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "DATASET", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-xplatform-dataset-index:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-xplatform-dataset-index@sha256:3dacff68fa10e013999fc940605d2f318d6422e941be392bea05ada8ca3d7908", "metadata": "operator.json", "name": "X 平台数据集索引", "operatorId": "operator-builtin-xplatform-dataset-index", @@ -18,6 +25,43 @@ "preferredSlug": "xplatform_dataset_index", "pythonImport": "backend.operators.builtin.xplatform_dataset_index.operator:run", "repositoryId": "builtin/xplatform_dataset_index", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_xplatform_index", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-xplatform-dataset-index", + "operatorSlug": "xplatform_dataset_index", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "xplatform-dataset-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "xplatform-dataset-files" ], diff --git a/backend/operators/builtin/xplatform_dataset_validate/repository.json b/backend/operators/builtin/xplatform_dataset_validate/repository.json index ba42d07d..56c4f555 100644 --- a/backend/operators/builtin/xplatform_dataset_validate/repository.json +++ b/backend/operators/builtin/xplatform_dataset_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "DATASET", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-xplatform-dataset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-xplatform-dataset-validate@sha256:10a3a8c6a775768205986dd9d81e938c5669be3ada995d03910cc28a3c58aaf3", "metadata": "operator.json", "name": "采集目录校验", "operatorId": "operator-builtin-xplatform-dataset-validate", @@ -18,6 +25,43 @@ "preferredSlug": "xplatform_dataset_validate", "pythonImport": "backend.operators.builtin.xplatform_dataset_validate.operator:run", "repositoryId": "builtin/xplatform_dataset_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_collection_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-xplatform-dataset-validate", + "operatorSlug": "xplatform_dataset_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "xplatform-dataset-index" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "xplatform-dataset-index" ], diff --git a/backend/operators/builtin/yolo_detection_validate/repository.json b/backend/operators/builtin/yolo_detection_validate/repository.json index 59767aa9..d6735cf5 100644 --- a/backend/operators/builtin/yolo_detection_validate/repository.json +++ b/backend/operators/builtin/yolo_detection_validate/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "DATASET", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-yolo-detection-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-yolo-detection-validate@sha256:2478bdf9dd3df22a9d5d6306c8ad0180891efc3d82bd028b50e5abfbc0e7783d", "metadata": "operator.json", "name": "YOLO 检测标注校验", "operatorId": "operator-builtin-yolo-detection-validate", @@ -18,6 +25,45 @@ "preferredSlug": "yolo_detection_validate", "pythonImport": "backend.operators.builtin.yolo_detection_validate.operator:run", "repositoryId": "builtin/yolo_detection_validate", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_yolo_detection_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-yolo-detection-validate", + "operatorSlug": "yolo_detection_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "yolo-label-files", + "image-files", + "pyyaml" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "yolo-label-files", "image-files", diff --git a/backend/operators/builtin/yolo_embodied_detection_train/operator.py b/backend/operators/builtin/yolo_embodied_detection_train/operator.py index e2c1c651..ee043f65 100644 --- a/backend/operators/builtin/yolo_embodied_detection_train/operator.py +++ b/backend/operators/builtin/yolo_embodied_detection_train/operator.py @@ -2,7 +2,7 @@ from __future__ import annotations from typing import Any -from backend.operators.builtin.command_model_operator import HelperMap, run_command_model_operator +from backend.operators.builtin.structured_model_operator import HelperMap, run_yolo_training def run( @@ -13,15 +13,11 @@ def run( inputs: list[dict[str, Any]], helpers: HelperMap, ) -> list[dict[str, Any]]: - return run_command_model_operator( + return run_yolo_training( run_id, node_id, output_dir, node_data, inputs, helpers, - operator_name="YOLO 具身目标检测训练", - model_family="YOLO detection", - recommended_gpu="1x NVIDIA A800 80GB", - recommended_gpu_memory_gb=24, ) diff --git a/backend/operators/builtin/yolo_embodied_detection_train/repository.json b/backend/operators/builtin/yolo_embodied_detection_train/repository.json index e988bb1c..668b2518 100644 --- a/backend/operators/builtin/yolo_embodied_detection_train/repository.json +++ b/backend/operators/builtin/yolo_embodied_detection_train/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "AI", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-yolo-embodied-detection-train:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-yolo-embodied-detection-train@sha256:52d3ddf838e032dc569a35cf1c937bd34241327a33e6f62efe91fadd1ea9a71a", "metadata": "operator.json", "name": "YOLO 具身目标检测训练", "operatorId": "operator-builtin-yolo-embodied-detection-train", @@ -18,6 +25,45 @@ "preferredSlug": "yolo_embodied_detection_train", "pythonImport": "backend.operators.builtin.yolo_embodied_detection_train.operator:run", "repositoryId": "builtin/yolo_embodied_detection_train", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": true, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_yolo_embodied_detection_train", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-yolo-embodied-detection-train", + "operatorSlug": "yolo_embodied_detection_train", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 1, + "recommendedGpuType": "NVIDIA A800 80GB", + "requiresGpu": true + }, + "runtimeDependencies": [ + "real-yolo-training-command", + "yolo-or-coco-dataset", + "gpu-a800" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "real-yolo-training-command", "yolo-or-coco-dataset", diff --git a/backend/operators/builtin/yolo_video_target_clip/repository.json b/backend/operators/builtin/yolo_video_target_clip/repository.json index 309159c1..8620d7de 100644 --- a/backend/operators/builtin/yolo_video_target_clip/repository.json +++ b/backend/operators/builtin/yolo_video_target_clip/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "VIDEO", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-yolo-video-target-clip:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-yolo-video-target-clip@sha256:36206dcc0ad5f80205273a0e983a66c4c47b65f2cecc94dba64d060016dafa67", "metadata": "operator.json", "name": "YOLO 视频目标检测裁剪", "operatorId": "operator-builtin-yolo-video-target-clip", @@ -18,6 +25,48 @@ "preferredSlug": "yolo_video_target_clip", "pythonImport": "backend.operators.builtin.yolo_video_target_clip.operator:run", "repositoryId": "builtin/yolo_video_target_clip", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": true, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_yolo_video_target_clip", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-yolo-video-target-clip", + "operatorSlug": "yolo_video_target_clip", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 1, + "recommendedGpuType": "NVIDIA A800 80GB", + "requiresGpu": true + }, + "runtimeDependencies": [ + "yolo-onnx-model", + "opencv-python-headless", + "numpy", + "ffmpeg", + "video-files", + "gpu-a800" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "yolo-onnx-model", "opencv-python-headless", diff --git a/backend/operators/builtin/zarr_trajectory_summary/repository.json b/backend/operators/builtin/zarr_trajectory_summary/repository.json index 214eb112..ddfb4c1c 100644 --- a/backend/operators/builtin/zarr_trajectory_summary/repository.json +++ b/backend/operators/builtin/zarr_trajectory_summary/repository.json @@ -9,8 +9,15 @@ "simulationOnlyAccepted": false }, "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "containerEntrypoint": "backend/operator-runner/main.py", "entrypoint": "operator.py", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-zarr-trajectory-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-zarr-trajectory-summary@sha256:10984c9423a77f6391516f96ad9dd6c708a1d7b0db80a87d5277165513fee505", "metadata": "operator.json", "name": "Zarr 轨迹数组概览", "operatorId": "operator-builtin-zarr-trajectory-summary", @@ -18,6 +25,44 @@ "preferredSlug": "zarr_trajectory_summary", "pythonImport": "backend.operators.builtin.zarr_trajectory_summary.operator:run", "repositoryId": "builtin/zarr_trajectory_summary", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_zarr_trajectory_summary", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-zarr-trajectory-summary", + "operatorSlug": "zarr_trajectory_summary", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "zarr-v2-v3-metadata", + "filesystem" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, "runtimeDependencies": [ "zarr-v2-v3-metadata", "filesystem" diff --git a/docs/operators/ACT-策略训练.md b/docs/operators/ACT-策略训练.md index 13a47f12..e903ac12 100644 --- a/docs/operators/ACT-策略训练.md +++ b/docs/operators/ACT-策略训练.md @@ -15,7 +15,7 @@ ## 平台语义 -具体 ACT 策略训练算子;平台负责 GPU 调度、上下游 artifact 注入和结果收集,训练本体必须由 COMMAND/SCRIPT 指向真实 ACT/LeRobot/自研训练代码,不接受无命令空跑。推荐 1 张 A800 80GB。 +ACT 策略训练算子;使用 DATASET_PATH/STATE_FIELD/ACTION_FIELD 等结构化参数读取真实轨迹样本,在独立 GPU 镜像中训练并输出 checkpoint.pt 与 metrics.json。推荐 1 张 A800 80GB。 ## 运行时契约 @@ -47,11 +47,14 @@ flowchart LR ## 关键超参数 -- `COMMAND`:类型 `string`,默认值 空字符串或由工作流显式填写。真实训练/推理命令;容器内执行,必须读取上游 INPUT_ARTIFACTS_PATH 或 SOURCE_PATH 并写出模型/报告文件 通常建议在工作流中显式配置。 -- `SCRIPT`:类型 `string`,默认值 空字符串或由工作流显式填写。可选脚本内容;与 COMMAND 二选一 通常建议在工作流中显式配置。 -- `SOURCE_PATH`:类型 `string`,默认值 空字符串或由工作流显式填写。可选真实数据集、模型或 manifest 路径 通常建议在工作流中显式配置。 -- `OUTPUT_MODEL_NAME`:类型 `string`,默认值 空字符串或由工作流显式填写。输出模型或权重名称 通常建议在工作流中显式配置。 -- `MODEL_FAMILY`:类型 `string`,默认值 `ACT`。模型族 可直接使用默认值,也可在节点级覆盖。 +- `DATASET_PATH`:类型 `string`,默认值 空字符串或由工作流显式填写。真实训练数据路径;支持 JSON/JSONL/CSV 或包含 samples.json/jsonl 的目录 通常建议在工作流中显式配置。 +- `STATE_FIELD`:类型 `string`,默认值 `state`。状态向量字段名 可直接使用默认值,也可在节点级覆盖。 +- `ACTION_FIELD`:类型 `string`,默认值 `action`。动作向量字段名 可直接使用默认值,也可在节点级覆盖。 +- `EPOCHS`:类型 `number`,默认值 `5`。训练轮数 可直接使用默认值,也可在节点级覆盖。 +- `BATCH_SIZE`:类型 `number`,默认值 `8`。训练 batch size 可直接使用默认值,也可在节点级覆盖。 +- `LEARNING_RATE`:类型 `number`,默认值 `0.001`。AdamW 学习率 可直接使用默认值,也可在节点级覆盖。 +- `HIDDEN_DIM`:类型 `number`,默认值 `128`。策略网络隐藏层宽度 可直接使用默认值,也可在节点级覆盖。 +- `CHUNK_SIZE`:类型 `number`,默认值 `1`。ACT action chunk size 可直接使用默认值,也可在节点级覆盖。 - `RECOMMENDED_GPU`:类型 `string`,默认值 `1x NVIDIA A800 80GB`。推荐 GPU 配置 可直接使用默认值,也可在节点级覆盖。 ## 执行前准备 @@ -78,11 +81,14 @@ flowchart LR "type": "builtin", "category": "AI", "hyperparams": { - "COMMAND": "<按节点填写>", - "SCRIPT": "<按节点填写>", - "SOURCE_PATH": "<按节点填写>", - "OUTPUT_MODEL_NAME": "<按节点填写>", - "MODEL_FAMILY": "ACT", + "DATASET_PATH": "<按节点填写>", + "STATE_FIELD": "state", + "ACTION_FIELD": "action", + "EPOCHS": "5", + "BATCH_SIZE": "8", + "LEARNING_RATE": "0.001", + "HIDDEN_DIM": "128", + "CHUNK_SIZE": "1", "RECOMMENDED_GPU": "1x NVIDIA A800 80GB" } } diff --git a/docs/operators/Diffusion-Policy-训练.md b/docs/operators/Diffusion-Policy-训练.md index ff920179..aae56b27 100644 --- a/docs/operators/Diffusion-Policy-训练.md +++ b/docs/operators/Diffusion-Policy-训练.md @@ -15,7 +15,7 @@ ## 平台语义 -具体 Diffusion Policy 训练算子;平台提供 GPU Job、artifact 和血缘,训练命令必须读取真实轨迹数据并产出 checkpoint/metrics。推荐 1 张 A800 80GB。 +Diffusion Policy 训练算子;使用 DATASET_PATH/STATE_FIELD/ACTION_FIELD/HORIZON/DIFFUSION_STEPS 等结构化参数读取真实轨迹数据,输出 checkpoint.pt 与 metrics.json。推荐 1 张 A800 80GB。 ## 运行时契约 @@ -47,11 +47,15 @@ flowchart LR ## 关键超参数 -- `COMMAND`:类型 `string`,默认值 空字符串或由工作流显式填写。真实训练/推理命令;容器内执行,必须读取上游 INPUT_ARTIFACTS_PATH 或 SOURCE_PATH 并写出模型/报告文件 通常建议在工作流中显式配置。 -- `SCRIPT`:类型 `string`,默认值 空字符串或由工作流显式填写。可选脚本内容;与 COMMAND 二选一 通常建议在工作流中显式配置。 -- `SOURCE_PATH`:类型 `string`,默认值 空字符串或由工作流显式填写。可选真实数据集、模型或 manifest 路径 通常建议在工作流中显式配置。 -- `OUTPUT_MODEL_NAME`:类型 `string`,默认值 空字符串或由工作流显式填写。输出模型或权重名称 通常建议在工作流中显式配置。 -- `MODEL_FAMILY`:类型 `string`,默认值 `Diffusion Policy`。模型族 可直接使用默认值,也可在节点级覆盖。 +- `DATASET_PATH`:类型 `string`,默认值 空字符串或由工作流显式填写。真实训练数据路径;支持 JSON/JSONL/CSV 或包含 samples.json/jsonl 的目录 通常建议在工作流中显式配置。 +- `STATE_FIELD`:类型 `string`,默认值 `state`。观测/状态向量字段名 可直接使用默认值,也可在节点级覆盖。 +- `ACTION_FIELD`:类型 `string`,默认值 `action`。动作向量字段名 可直接使用默认值,也可在节点级覆盖。 +- `EPOCHS`:类型 `number`,默认值 `5`。训练轮数 可直接使用默认值,也可在节点级覆盖。 +- `BATCH_SIZE`:类型 `number`,默认值 `8`。训练 batch size 可直接使用默认值,也可在节点级覆盖。 +- `LEARNING_RATE`:类型 `number`,默认值 `0.001`。AdamW 学习率 可直接使用默认值,也可在节点级覆盖。 +- `HIDDEN_DIM`:类型 `number`,默认值 `256`。策略网络隐藏层宽度 可直接使用默认值,也可在节点级覆盖。 +- `HORIZON`:类型 `number`,默认值 `16`。预测 horizon 可直接使用默认值,也可在节点级覆盖。 +- `DIFFUSION_STEPS`:类型 `number`,默认值 `32`。扩散步数配置,用于训练记录和下游复现 可直接使用默认值,也可在节点级覆盖。 - `RECOMMENDED_GPU`:类型 `string`,默认值 `1x NVIDIA A800 80GB`。推荐 GPU 配置 可直接使用默认值,也可在节点级覆盖。 ## 执行前准备 @@ -78,11 +82,15 @@ flowchart LR "type": "builtin", "category": "AI", "hyperparams": { - "COMMAND": "<按节点填写>", - "SCRIPT": "<按节点填写>", - "SOURCE_PATH": "<按节点填写>", - "OUTPUT_MODEL_NAME": "<按节点填写>", - "MODEL_FAMILY": "Diffusion Policy", + "DATASET_PATH": "<按节点填写>", + "STATE_FIELD": "state", + "ACTION_FIELD": "action", + "EPOCHS": "5", + "BATCH_SIZE": "8", + "LEARNING_RATE": "0.001", + "HIDDEN_DIM": "256", + "HORIZON": "16", + "DIFFUSION_STEPS": "32", "RECOMMENDED_GPU": "1x NVIDIA A800 80GB" } } diff --git a/docs/operators/OpenVLA-微调训练.md b/docs/operators/OpenVLA-微调训练.md index 8f786851..d2516509 100644 --- a/docs/operators/OpenVLA-微调训练.md +++ b/docs/operators/OpenVLA-微调训练.md @@ -15,7 +15,7 @@ ## 平台语义 -具体 OpenVLA 微调算子;推荐先用 LoRA/QLoRA 在 1 张 A800 80GB 上训练,full fine-tune 可在命令中改为多卡。必须提供真实权重、数据和训练命令。 +OpenVLA 微调算子;使用 BASE_MODEL、DATASET_PATH、TEXT_FIELD、TARGET_FIELD 等结构化参数,通过 Transformers/PyTorch 对真实指令-动作样本训练并输出模型目录与 metrics.json。推荐 1 张 A800 80GB。 ## 运行时契约 @@ -47,12 +47,15 @@ flowchart LR ## 关键超参数 -- `COMMAND`:类型 `string`,默认值 空字符串或由工作流显式填写。真实训练/推理命令;容器内执行,必须读取上游 INPUT_ARTIFACTS_PATH 或 SOURCE_PATH 并写出模型/报告文件 通常建议在工作流中显式配置。 -- `SCRIPT`:类型 `string`,默认值 空字符串或由工作流显式填写。可选脚本内容;与 COMMAND 二选一 通常建议在工作流中显式配置。 -- `SOURCE_PATH`:类型 `string`,默认值 空字符串或由工作流显式填写。可选真实数据集、模型或 manifest 路径 通常建议在工作流中显式配置。 -- `OUTPUT_MODEL_NAME`:类型 `string`,默认值 空字符串或由工作流显式填写。输出模型或权重名称 通常建议在工作流中显式配置。 -- `BASE_MODEL`:类型 `string`,默认值 `openvla/openvla-7b`。基础模型或本地权重路径 可直接使用默认值,也可在节点级覆盖。 -- `FINETUNE_METHOD`:类型 `string`,默认值 `lora`。lora、qlora 或 full 可直接使用默认值,也可在节点级覆盖。 +- `BASE_MODEL`:类型 `string`,默认值 空字符串或由工作流显式填写。Hugging Face 或本地基础模型路径 通常建议在工作流中显式配置。 +- `DATASET_PATH`:类型 `string`,默认值 空字符串或由工作流显式填写。真实微调数据 JSON/JSONL/CSV 路径 通常建议在工作流中显式配置。 +- `TEXT_FIELD`:类型 `string`,默认值 `instruction`。指令/观测文本字段名 可直接使用默认值,也可在节点级覆盖。 +- `TARGET_FIELD`:类型 `string`,默认值 `action`。目标动作文本字段名 可直接使用默认值,也可在节点级覆盖。 +- `MAX_LENGTH`:类型 `number`,默认值 `256`。token 最大长度 可直接使用默认值,也可在节点级覆盖。 +- `EPOCHS`:类型 `number`,默认值 `1`。训练轮数 可直接使用默认值,也可在节点级覆盖。 +- `BATCH_SIZE`:类型 `number`,默认值 `1`。训练 batch size 可直接使用默认值,也可在节点级覆盖。 +- `LEARNING_RATE`:类型 `number`,默认值 `0.00005`。AdamW 学习率 可直接使用默认值,也可在节点级覆盖。 +- `PRECISION`:类型 `string`,默认值 `bf16`。推荐精度配置 可直接使用默认值,也可在节点级覆盖。 - `RECOMMENDED_GPU`:类型 `string`,默认值 `1x NVIDIA A800 80GB`。推荐 GPU 配置 可直接使用默认值,也可在节点级覆盖。 ## 执行前准备 @@ -79,12 +82,15 @@ flowchart LR "type": "builtin", "category": "AI", "hyperparams": { - "COMMAND": "<按节点填写>", - "SCRIPT": "<按节点填写>", - "SOURCE_PATH": "<按节点填写>", - "OUTPUT_MODEL_NAME": "<按节点填写>", - "BASE_MODEL": "openvla/openvla-7b", - "FINETUNE_METHOD": "lora", + "BASE_MODEL": "<按节点填写>", + "DATASET_PATH": "<按节点填写>", + "TEXT_FIELD": "instruction", + "TARGET_FIELD": "action", + "MAX_LENGTH": "256", + "EPOCHS": "1", + "BATCH_SIZE": "1", + "LEARNING_RATE": "0.00005", + "PRECISION": "bf16", "RECOMMENDED_GPU": "1x NVIDIA A800 80GB" } } diff --git a/docs/operators/OpenVLA-推理.md b/docs/operators/OpenVLA-推理.md index 6a4ff17b..7b965bfd 100644 --- a/docs/operators/OpenVLA-推理.md +++ b/docs/operators/OpenVLA-推理.md @@ -15,7 +15,7 @@ ## 平台语义 -具体 OpenVLA 推理算子;用于替代笼统模型推理,必须由 COMMAND/SCRIPT 调用真实模型权重并生成预测 artifact。推荐 1 张 A800,量化模型可降低显存。 +OpenVLA 推理算子;使用 MODEL_PATH 或 BASE_MODEL 加载真实 Transformers 权重,读取 PROMPT 或上游样本 instruction,并输出 inference-result.json。推荐 1 张 A800,量化模型可降低显存。 ## 运行时契约 @@ -47,12 +47,12 @@ flowchart LR ## 关键超参数 -- `COMMAND`:类型 `string`,默认值 空字符串或由工作流显式填写。真实训练/推理命令;容器内执行,必须读取上游 INPUT_ARTIFACTS_PATH 或 SOURCE_PATH 并写出模型/报告文件 通常建议在工作流中显式配置。 -- `SCRIPT`:类型 `string`,默认值 空字符串或由工作流显式填写。可选脚本内容;与 COMMAND 二选一 通常建议在工作流中显式配置。 -- `SOURCE_PATH`:类型 `string`,默认值 空字符串或由工作流显式填写。可选真实数据集、模型或 manifest 路径 通常建议在工作流中显式配置。 -- `OUTPUT_MODEL_NAME`:类型 `string`,默认值 空字符串或由工作流显式填写。输出模型或权重名称 通常建议在工作流中显式配置。 -- `MODEL_PATH`:类型 `string`,默认值 空字符串或由工作流显式填写。OpenVLA 权重或 adapter 路径 通常建议在工作流中显式配置。 -- `PRECISION`:类型 `string`,默认值 `bf16`。bf16、fp16、int8 或 int4 可直接使用默认值,也可在节点级覆盖。 +- `MODEL_PATH`:类型 `string`,默认值 空字符串或由工作流显式填写。本地模型目录或 Hugging Face 模型名 通常建议在工作流中显式配置。 +- `BASE_MODEL`:类型 `string`,默认值 空字符串或由工作流显式填写。未提供 MODEL_PATH 时使用的基础模型 通常建议在工作流中显式配置。 +- `PROMPT`:类型 `string`,默认值 空字符串或由工作流显式填写。推理 prompt;为空则从上游样本读取 TEXT_FIELD 通常建议在工作流中显式配置。 +- `TEXT_FIELD`:类型 `string`,默认值 `instruction`。上游样本指令字段名 可直接使用默认值,也可在节点级覆盖。 +- `MAX_NEW_TOKENS`:类型 `number`,默认值 `64`。生成 token 数 可直接使用默认值,也可在节点级覆盖。 +- `PRECISION`:类型 `string`,默认值 `bf16`。推荐精度配置 可直接使用默认值,也可在节点级覆盖。 - `RECOMMENDED_GPU`:类型 `string`,默认值 `1x NVIDIA A800 80GB`。推荐 GPU 配置 可直接使用默认值,也可在节点级覆盖。 ## 执行前准备 @@ -79,11 +79,11 @@ flowchart LR "type": "builtin", "category": "AI", "hyperparams": { - "COMMAND": "<按节点填写>", - "SCRIPT": "<按节点填写>", - "SOURCE_PATH": "<按节点填写>", - "OUTPUT_MODEL_NAME": "<按节点填写>", "MODEL_PATH": "<按节点填写>", + "BASE_MODEL": "<按节点填写>", + "PROMPT": "<按节点填写>", + "TEXT_FIELD": "instruction", + "MAX_NEW_TOKENS": "64", "PRECISION": "bf16", "RECOMMENDED_GPU": "1x NVIDIA A800 80GB" } diff --git a/docs/operators/YOLO-具身目标检测训练.md b/docs/operators/YOLO-具身目标检测训练.md index f5d2aa51..36585bdd 100644 --- a/docs/operators/YOLO-具身目标检测训练.md +++ b/docs/operators/YOLO-具身目标检测训练.md @@ -15,7 +15,7 @@ ## 平台语义 -具体 YOLO 检测训练算子;必须读取真实 YOLO/COCO 数据集并输出权重、metrics 和类别配置。A800 对 YOLOv8/YOLOv10/YOLO11 中大模型训练充足,默认请求 1 卡。 +YOLO 具身目标检测训练算子;使用 Ultralytics API 读取真实 YOLO/COCO data.yaml,输出 best/last checkpoint 与 metrics.json。A800 对 YOLOv8/YOLOv10/YOLO11 中大模型训练充足,默认请求 1 卡。 ## 运行时契约 @@ -47,12 +47,13 @@ flowchart LR ## 关键超参数 -- `COMMAND`:类型 `string`,默认值 空字符串或由工作流显式填写。真实训练/推理命令;容器内执行,必须读取上游 INPUT_ARTIFACTS_PATH 或 SOURCE_PATH 并写出模型/报告文件 通常建议在工作流中显式配置。 -- `SCRIPT`:类型 `string`,默认值 空字符串或由工作流显式填写。可选脚本内容;与 COMMAND 二选一 通常建议在工作流中显式配置。 -- `SOURCE_PATH`:类型 `string`,默认值 空字符串或由工作流显式填写。可选真实数据集、模型或 manifest 路径 通常建议在工作流中显式配置。 -- `OUTPUT_MODEL_NAME`:类型 `string`,默认值 空字符串或由工作流显式填写。输出模型或权重名称 通常建议在工作流中显式配置。 -- `MODEL_SIZE`:类型 `string`,默认值 `yolo11m`。建议模型规格 可直接使用默认值,也可在节点级覆盖。 -- `DATA_YAML`:类型 `string`,默认值 空字符串或由工作流显式填写。YOLO data.yaml 路径 通常建议在工作流中显式配置。 +- `DATA_YAML`:类型 `string`,默认值 空字符串或由工作流显式填写。真实 YOLO/COCO data.yaml 路径 通常建议在工作流中显式配置。 +- `MODEL_SIZE`:类型 `string`,默认值 `yolo11n.pt`。Ultralytics 模型名或权重路径 可直接使用默认值,也可在节点级覆盖。 +- `EPOCHS`:类型 `number`,默认值 `10`。训练轮数 可直接使用默认值,也可在节点级覆盖。 +- `IMAGE_SIZE`:类型 `number`,默认值 `640`。输入图片尺寸 可直接使用默认值,也可在节点级覆盖。 +- `BATCH_SIZE`:类型 `number`,默认值 `8`。训练 batch size 可直接使用默认值,也可在节点级覆盖。 +- `CONF`:类型 `number`,默认值 `0.25`。后续检测推荐置信度 可直接使用默认值,也可在节点级覆盖。 +- `TARGET_LABELS`:类型 `string`,默认值 空字符串或由工作流显式填写。关注类别,逗号分隔 通常建议在工作流中显式配置。 - `RECOMMENDED_GPU`:类型 `string`,默认值 `1x NVIDIA A800 80GB`。推荐 GPU 配置 可直接使用默认值,也可在节点级覆盖。 ## 执行前准备 @@ -79,12 +80,13 @@ flowchart LR "type": "builtin", "category": "AI", "hyperparams": { - "COMMAND": "<按节点填写>", - "SCRIPT": "<按节点填写>", - "SOURCE_PATH": "<按节点填写>", - "OUTPUT_MODEL_NAME": "<按节点填写>", - "MODEL_SIZE": "yolo11m", "DATA_YAML": "<按节点填写>", + "MODEL_SIZE": "yolo11n.pt", + "EPOCHS": "10", + "IMAGE_SIZE": "640", + "BATCH_SIZE": "8", + "CONF": "0.25", + "TARGET_LABELS": "<按节点填写>", "RECOMMENDED_GPU": "1x NVIDIA A800 80GB" } } diff --git a/docs/operators/模型推理.md b/docs/operators/模型推理.md index 9c5b05c8..f0beb997 100644 --- a/docs/operators/模型推理.md +++ b/docs/operators/模型推理.md @@ -15,7 +15,7 @@ ## 平台语义 -执行真实推理节点;优先执行 COMMAND/SCRIPT 后端,否则读取 MODEL_JSON 或 MODEL_PATH,并支持 rule、linear、nearest_centroid 三类可解释轻量模型;拒绝 constant 等硬编码输出模型。 +执行结构化模型推理节点;读取 MODEL_JSON 或 MODEL_PATH,支持 rule、linear、nearest_centroid 三类可解释模型;自定义命令请使用 Shell 命令算子,constant 等硬编码输出模型会被拒绝。 ## 运行时契约 @@ -51,8 +51,6 @@ flowchart LR - `MODEL_JSON`:类型 `json`,默认值 空字符串或由工作流显式填写。模型 JSON,支持 rule/linear/nearest_centroid;constant 等硬编码输出模型会被拒绝 通常建议在工作流中显式配置。 - `MODEL_PATH`:类型 `string`,默认值 空字符串或由工作流显式填写。operator-runner 容器内可读模型 JSON 文件路径 通常建议在工作流中显式配置。 - `INPUT_JSON`:类型 `json`,默认值 空字符串或由工作流显式填写。可选内联输入 JSON;为空时使用上游 artifact 通常建议在工作流中显式配置。 -- `COMMAND`:类型 `string`,默认值 空字符串或由工作流显式填写。容器内执行的命令 通常建议在工作流中显式配置。 -- `SCRIPT`:类型 `string`,默认值 空字符串或由工作流显式填写。可执行脚本内容 通常建议在工作流中显式配置。 ## 执行前准备 @@ -80,9 +78,7 @@ flowchart LR "hyperparams": { "MODEL_JSON": "<按节点填写>", "MODEL_PATH": "<按节点填写>", - "INPUT_JSON": "<按节点填写>", - "COMMAND": "<按节点填写>", - "SCRIPT": "<按节点填写>" + "INPUT_JSON": "<按节点填写>" } } ``` diff --git a/internal/api/builtin_operators_catalog.json b/internal/api/builtin_operators_catalog.json index 6dce4381..cc8a365e 100644 --- a/internal/api/builtin_operators_catalog.json +++ b/internal/api/builtin_operators_catalog.json @@ -1,5819 +1,10240 @@ [ { - "id": "operator-builtin-real-source", - "name": "真实数据源", - "type": "builtin", "category": "IO", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "description": "从 SOURCE_TEXT、SOURCE_JSON、SOURCE_PATH 或上游输入生成真实源数据清单。", - "outputs": [ - { - "id": "out-1", - "name": "manifest", - "type": "json", - "description": "源数据清单", - "transmission": "bus" - } - ], - "skillDefinition": "工作流起点。读取真实输入源并生成结构化清单,供下游节点消费。", "hyperparams": [ { - "key": "DATA_SOURCE", - "value": "", "description": "留空自动推断;可显式设置 INLINE、LOCAL、BOS 或 UPSTREAM", - "type": "string" + "key": "DATA_SOURCE", + "type": "string", + "value": "" }, { - "key": "SOURCE_TEXT", - "value": "", "description": "直接注入文本内容", - "type": "string" + "key": "SOURCE_TEXT", + "type": "string", + "value": "" }, { - "key": "SOURCE_JSON", - "value": "", "description": "直接注入 JSON 内容", - "type": "string" + "key": "SOURCE_JSON", + "type": "string", + "value": "" }, { - "key": "SOURCE_PATH", - "value": "", "description": "从路径读取真实数据", - "type": "string" + "key": "SOURCE_PATH", + "type": "string", + "value": "" }, { - "key": "PAYLOAD_OUTPUT_ID", - "value": "", "description": "可选:SOURCE_TEXT 或 SOURCE_JSON 内联输入额外写出 payload artifact 时绑定的输出口 id/name", - "type": "string" + "key": "PAYLOAD_OUTPUT_ID", + "type": "string", + "value": "" }, { - "key": "OUTPUT_NAME", - "value": "", "description": "可选:内联 payload artifact 文件名,例如 payload.json、report.csv 或 notes.txt", - "type": "string" + "key": "OUTPUT_NAME", + "type": "string", + "value": "" } ], - "runtimeSource": "backend/operators/builtin/real_source/operator.py", - "runtimeHandler": "builtin_source" - }, - { - "id": "operator-builtin-bos-incremental-scan", - "name": "BOS 增量扫描", - "type": "builtin", - "category": "IO", - "description": "按 bucket/prefix 增量扫描 BOS,仅输出本次新增对象并提交 checkpoint 给 runtime。", + "id": "operator-builtin-real-source", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-real-source:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-real-source@sha256:833d6030b8c10cabfddb3cad54137bb08f232f2bf5071863021276e6b6311d66", + "name": "真实数据源", "outputs": [ { + "description": "源数据清单", "id": "out-1", "name": "manifest", - "type": "json", - "description": "增量对象清单", - "transmission": "bus" + "transmission": "bus", + "type": "json" } ], - "skillDefinition": "统一工作流里的有状态源算子。每次 run 只产出 BOS 中新增的对象,状态由 runtime 在 run 成功后提交。", - "hyperparams": [ - { - "key": "DATA_SOURCE", - "value": "BOS", - "description": "固定为 BOS", - "type": "string" + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true }, - { - "key": "SOURCE_PATH", - "value": "", - "description": "完整 bos://bucket/prefix 路径", - "type": "string" + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true }, - { - "key": "BOS_ENDPOINT", - "value": "", - "description": "BOS Endpoint;留空则依赖全局 BOS_ENDPOINT,若全局也为空会在执行时报错", - "type": "string" + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_source", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true }, - { - "key": "BOS_ACCESS_KEY_ID", - "value": "", - "description": "BOS Access Key ID", - "type": "string" + "operatorId": "operator-builtin-real-source", + "operatorSlug": "real_source", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true }, - { - "key": "BOS_SECRET_ACCESS_KEY", - "value": "", - "description": "BOS Secret Access Key", - "type": "string" + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false }, - { - "key": "SCAN_MODE", - "value": "INCREMENTAL", - "description": "固定为 INCREMENTAL", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/bos_incremental_scan/operator.py", - "runtimeHandler": "builtin_source" + "runtimeDependencies": [ + "filesystem" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_source", + "runtimeSource": "backend/operators/builtin/real_source/operator.py", + "skillDefinition": "工作流起点。读取真实输入源并生成结构化清单,供下游节点消费。", + "type": "builtin" + }, + { + "category": "IO", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "按 bucket/prefix 增量扫描 BOS,仅输出本次新增对象并提交 checkpoint 给 runtime。", + "hyperparams": [ + { + "description": "固定为 BOS", + "key": "DATA_SOURCE", + "type": "string", + "value": "BOS" + }, + { + "description": "完整 bos://bucket/prefix 路径", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "BOS Endpoint;留空则依赖全局 BOS_ENDPOINT,若全局也为空会在执行时报错", + "key": "BOS_ENDPOINT", + "type": "string", + "value": "" + }, + { + "description": "BOS Access Key ID", + "key": "BOS_ACCESS_KEY_ID", + "type": "string", + "value": "" + }, + { + "description": "BOS Secret Access Key", + "key": "BOS_SECRET_ACCESS_KEY", + "type": "string", + "value": "" + }, + { + "description": "固定为 INCREMENTAL", + "key": "SCAN_MODE", + "type": "string", + "value": "INCREMENTAL" + } + ], + "id": "operator-builtin-bos-incremental-scan", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-bos-incremental-scan:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-bos-incremental-scan@sha256:9077ae0f63cdf0fe76ce363ea08b84f071b2fc3c0ef516bd797a8d63eba91249", + "name": "BOS 增量扫描", + "outputs": [ + { + "description": "增量对象清单", + "id": "out-1", + "name": "manifest", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_source", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-bos-incremental-scan", + "operatorSlug": "bos_incremental_scan", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "baidu-bos" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_source", + "runtimeSource": "backend/operators/builtin/bos_incremental_scan/operator.py", + "skillDefinition": "统一工作流里的有状态源算子。每次 run 只产出 BOS 中新增的对象,状态由 runtime 在 run 成功后提交。", + "type": "builtin" }, { - "id": "operator-builtin-model-inference", - "name": "模型推理", - "type": "builtin", "category": "AI", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "description": "执行真实本地模型 JSON 推理、命令后端推理或模型镜像推理。", + "hyperparams": [ + { + "description": "模型 JSON,支持 rule/linear/nearest_centroid;constant 等硬编码输出模型会被拒绝", + "key": "MODEL_JSON", + "type": "json", + "value": "" + }, + { + "description": "operator-runner 容器内可读模型 JSON 文件路径", + "key": "MODEL_PATH", + "type": "string", + "value": "" + }, + { + "description": "可选内联输入 JSON;为空时使用上游 artifact", + "key": "INPUT_JSON", + "type": "json", + "value": "" + } + ], + "id": "operator-builtin-model-inference", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-model-inference:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-model-inference@sha256:a31cf2777cda7b213f4b3fa7e004ed5ddcd31fe5f3503c5cbb039be1f6bfaaa4", "inputs": [ { + "description": "输入数据", "id": "in-1", "name": "input", - "type": "json", - "description": "输入数据", - "transmission": "bus" + "transmission": "bus", + "type": "json" } ], + "name": "模型推理", "outputs": [ { + "description": "推理结果", "id": "out-1", "name": "result", - "type": "json", - "description": "推理结果", - "transmission": "bus" + "transmission": "bus", + "type": "json" } ], - "skillDefinition": "执行真实推理节点;优先执行 COMMAND/SCRIPT 后端,否则读取 MODEL_JSON 或 MODEL_PATH,并支持 rule、linear、nearest_centroid 三类可解释轻量模型;拒绝 constant 等硬编码输出模型。", - "hyperparams": [ - { - "key": "MODEL_JSON", - "value": "", - "description": "模型 JSON,支持 rule/linear/nearest_centroid;constant 等硬编码输出模型会被拒绝", - "type": "json" + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true }, - { - "key": "MODEL_PATH", - "value": "", - "description": "operator-runner 容器内可读模型 JSON 文件路径", - "type": "string" + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true }, - { - "key": "INPUT_JSON", - "value": "", - "description": "可选内联输入 JSON;为空时使用上游 artifact", - "type": "json" + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_model_inference", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true }, - { - "key": "COMMAND", - "value": "", - "description": "容器内执行的命令", - "type": "string" + "operatorId": "operator-builtin-model-inference", + "operatorSlug": "model_inference", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true }, - { - "key": "SCRIPT", - "value": "", - "description": "可执行脚本内容", - "type": "string" - } - ], + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-model-or-command" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_model_inference", "runtimeSource": "backend/operators/builtin/model_inference/operator.py", - "runtimeHandler": "builtin_model_inference" + "skillDefinition": "执行结构化模型推理节点;读取 MODEL_JSON 或 MODEL_PATH,支持 rule、linear、nearest_centroid 三类可解释模型;自定义命令请使用 Shell 命令算子,constant 等硬编码输出模型会被拒绝。", + "type": "builtin" }, { - "id": "operator-builtin-merge", - "name": "数据合并", - "type": "builtin", "category": "MATH", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "description": "将多个 JSON 输入合并成统一结果。", + "id": "operator-builtin-merge", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-merge:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-merge@sha256:43dd9f250d94e9dcc2af193aa03a602a3f27ce4d3b119f5063760638cd1db0d9", "inputs": [ { + "description": "左输入", "id": "in-1", "name": "left", - "type": "json", - "description": "左输入", - "transmission": "bus" - }, - { - "id": "in-2", - "name": "right", - "type": "json", - "description": "右输入", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "merged", - "type": "json", - "description": "合并结果", - "transmission": "bus" - } - ], - "skillDefinition": "对多个 JSON 输入执行真实合并。", - "runtimeSource": "backend/operators/builtin/merge/operator.py", - "runtimeHandler": "builtin_merge" - }, - { - "id": "operator-builtin-upload", - "name": "结果上传", - "type": "builtin", - "category": "IO", - "description": "收集上游真实输出并生成上传结果清单。", - "inputs": [ - { - "id": "in-1", - "name": "result", - "type": "json", - "description": "待上传结果", - "transmission": "bus" - } - ], - "skillDefinition": "复制或汇总上游产物,生成上传结果清单。", - "runtimeSource": "backend/operators/builtin/upload/operator.py", - "runtimeHandler": "builtin_upload" - }, - { - "id": "operator-builtin-human-review", - "name": "人工审核", - "type": "builtin", - "category": "HUMAN", - "description": "将真实输入送入人工审核并等待审核动作恢复流程。", - "inputs": [ - { - "id": "in-1", - "name": "candidate", - "type": "json", - "description": "候选数据", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "approved", - "name": "approved", - "type": "json", - "description": "审核通过", - "transmission": "bus" - }, - { - "id": "rejected", - "name": "rejected", - "type": "json", - "description": "审核拒绝", - "transmission": "bus" - } - ], - "skillDefinition": "基于真实输入生成审核任务并等待人工处理。", - "runtimeSource": "backend/operators/builtin/human_review/operator.py", - "runtimeHandler": "builtin_human_review" - }, - { - "id": "operator-builtin-directory-scan", - "name": "目录扫描", - "type": "builtin", - "category": "IO", - "description": "对真实目录做文件统计、扩展名分布和样例文件扫描。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "目录输入", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "summary", - "type": "json", - "description": "目录概览", - "transmission": "bus" - } - ], - "skillDefinition": "读取 SOURCE_PATH 或上游 source manifest,输出真实目录概览。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "待扫描目录", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/directory_scan/operator.py", - "runtimeHandler": "builtin_directory_scan" - }, - { - "id": "operator-builtin-csv-to-json", - "name": "CSV 转 JSON", - "type": "builtin", - "category": "DATA", - "description": "将真实 CSV/TSV 文件转换成结构化 JSON 行列表。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "CSV/TSV 输入", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "table", - "type": "json", - "description": "JSON 行列表", - "transmission": "bus" - } - ], - "skillDefinition": "读取 SOURCE_PATH 或上游 source manifest 指向的 csv/tsv 文件,输出完整 JSON 行数据。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "csv/tsv 文件路径", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/csv_to_json/operator.py", - "runtimeHandler": "builtin_csv_to_json" - }, - { - "id": "operator-builtin-json-to-csv", - "name": "JSON 转 CSV", - "type": "builtin", - "category": "DATA", - "description": "将 JSON 行数据导出为真实 CSV 文件。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "JSON 行列表", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "csv", - "type": "file", - "description": "CSV 文件", - "transmission": "bus" - } - ], - "skillDefinition": "把上游 JSON items/entries 或单条对象导出为真实 CSV 文件,并附带摘要。", - "runtimeSource": "backend/operators/builtin/json_to_csv/operator.py", - "runtimeHandler": "builtin_json_to_csv" - }, - { - "id": "operator-builtin-json-field-extract", - "name": "JSON 字段提取", - "type": "builtin", - "category": "DATA", - "description": "按字段路径从 JSON 输入中提取指定值。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "JSON 输入", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "extracted", - "type": "json", - "description": "提取结果", - "transmission": "bus" - } - ], - "skillDefinition": "根据 FIELD_PATH 从 JSON 输入中提取字段。", - "hyperparams": [ - { - "key": "FIELD_PATH", - "value": "", - "description": "点分字段路径,例如 metrics.score", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/json_field_extract/operator.py", - "runtimeHandler": "builtin_json_extract" - }, - { - "id": "operator-builtin-json-filter", - "name": "JSON 过滤", - "type": "builtin", - "category": "DATA", - "description": "按字段条件过滤 JSON 列表。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "JSON 输入", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "filtered", - "type": "json", - "description": "过滤结果", - "transmission": "bus" - } - ], - "skillDefinition": "按 FIELD_PATH + OPERATOR + EXPECTED 条件真实过滤 JSON items。", - "hyperparams": [ - { - "key": "FIELD_PATH", - "value": "", - "description": "待判断字段路径", - "type": "string" - }, - { - "key": "OPERATOR", - "value": "eq", - "description": "eq/ne/contains/gt/gte/lt/lte/exists/missing/in;显式覆盖时必须是受支持值", - "type": "string" - }, - { - "key": "EXPECTED", - "value": "", - "description": "字符串比较值", - "type": "string" - }, - { - "key": "EXPECTED_JSON", - "value": "", - "description": "JSON 格式比较值,优先于 EXPECTED", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/json_filter/operator.py", - "runtimeHandler": "builtin_json_filter" - }, - { - "id": "operator-builtin-json-map-rename", - "name": "JSON 映射重命名", - "type": "builtin", - "category": "DATA", - "description": "按映射表批量重命名 JSON 字段。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "JSON 输入", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "mapped", - "type": "json", - "description": "映射结果", - "transmission": "bus" - } - ], - "skillDefinition": "根据 FIELD_MAP_JSON 真实执行字段名重映射;源字段和目标字段必须是非空字符串,且不能把多个输入字段写到同一输出字段。", - "hyperparams": [ - { - "key": "FIELD_MAP_JSON", - "value": "{}", - "description": "JSON 字段映射,例如 {\"old\":\"new\"};源/目标字段必须是非空字符串且不得导致输出字段覆盖", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/json_map_rename/operator.py", - "runtimeHandler": "builtin_json_map" - }, - { - "id": "operator-builtin-template-render", - "name": "文本模板渲染", - "type": "builtin", - "category": "UTIL", - "description": "使用上游 JSON 上下文渲染真实文本模板文件。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "JSON 输入", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "rendered", - "type": "file", - "description": "渲染后的文本文件", - "transmission": "bus" - } - ], - "skillDefinition": "支持 {{field}} 和 Python format 风格占位符,输出真实文本文件。", - "hyperparams": [ - { - "key": "TEMPLATE_TEXT", - "value": "", - "description": "模板文本", - "type": "string" - }, - { - "key": "OUTPUT_NAME", - "value": "rendered.txt", - "description": "输出文件名", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/template_render/operator.py", - "runtimeHandler": "builtin_text_template" - }, - { - "id": "operator-builtin-conditional-route", - "name": "条件分流", - "type": "builtin", - "category": "LOGIC", - "description": "按字段条件将输入拆分到 matched / unmatched 两个真实输出端口。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "JSON 输入", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "matched", - "name": "matched", - "type": "json", - "description": "命中规则的数据", - "transmission": "bus" - }, - { - "id": "unmatched", - "name": "unmatched", - "type": "json", - "description": "未命中规则的数据", - "transmission": "bus" - } - ], - "skillDefinition": "按 FIELD_PATH + OPERATOR + EXPECTED 条件把真实数据分流到两个输出端口。", - "hyperparams": [ - { - "key": "FIELD_PATH", - "value": "", - "description": "待判断字段路径", - "type": "string" - }, - { - "key": "OPERATOR", - "value": "eq", - "description": "eq/ne/contains/gt/gte/lt/lte/exists/missing/in;显式覆盖时必须是受支持值", - "type": "string" - }, - { - "key": "EXPECTED", - "value": "", - "description": "字符串比较值", - "type": "string" - }, - { - "key": "EXPECTED_JSON", - "value": "", - "description": "JSON 格式比较值,优先于 EXPECTED", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/conditional_route/operator.py", - "runtimeHandler": "builtin_condition_router" - }, - { - "id": "operator-builtin-xplatform-dataset-index", - "name": "X 平台数据集索引", - "type": "builtin", - "category": "DATASET", - "description": "扫描 X-platform 采集目录,解析 meta.json、video_meta.json 和视频文件索引。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "目录输入", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "dataset", - "type": "json", - "description": "数据集索引", - "transmission": "bus" - } - ], - "skillDefinition": "面向 BJ_***/DJI_***/meta.json/video_meta.json 结构的真实数据集索引算子,输出采集目录和视频清单。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "X-platform 根目录或单个采集目录", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/xplatform_dataset_index/operator.py", - "runtimeHandler": "builtin_xplatform_index" - }, - { - "id": "operator-builtin-xplatform-dataset-validate", - "name": "采集目录校验", - "type": "builtin", - "category": "DATASET", - "description": "校验采集目录是否包含 meta.json、video_meta.json,以及可选 intrinsics.json。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "数据集索引", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validated", - "type": "json", - "description": "校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "对 X-platform 数据集索引结果执行真实结构校验,输出每个采集目录的有效性报告。", - "hyperparams": [ - { - "key": "REQUIRE_INTRINSICS", - "value": "false", - "description": "是否要求 intrinsics.json 必须存在", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/xplatform_dataset_validate/operator.py", - "runtimeHandler": "builtin_collection_validate" - }, - { - "id": "operator-builtin-video-quality", - "name": "视频质检", - "type": "builtin", - "category": "VIDEO", - "description": "基于真实视频帧计算清晰度、抖动度和无手率估计。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "视频清单", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "quality", - "type": "json", - "description": "质检结果", - "transmission": "bus" - } - ], - "skillDefinition": "使用 OpenCV 对真实 mp4 做质检分析,输出 clarity / shake / handOcclusion / passed。", - "hyperparams": [ - { - "key": "MIN_CLARITY", - "value": "0.8", - "description": "清晰度最低阈值", - "type": "string" - }, - { - "key": "MAX_SHAKE", - "value": "0.1", - "description": "抖动度最高阈值", - "type": "string" - }, - { - "key": "MAX_HAND_OCCLUSION", - "value": "0.1", - "description": "无手率最高阈值", - "type": "string" - }, - { - "key": "CLARITY_SCALE", - "value": "35", - "description": "清晰度归一化尺度,显式覆盖时必须为大于 0 的有限数字", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/video_quality/operator.py", - "runtimeHandler": "builtin_video_quality" - }, - { - "id": "operator-builtin-video-filter", - "name": "数据过滤", - "type": "builtin", - "category": "VIDEO", - "description": "按日期、地点、场景和质检阈值过滤有效视频。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "视频与元数据输入", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "filtered", - "type": "json", - "description": "过滤结果", - "transmission": "bus" - } - ], - "skillDefinition": "针对 X-platform 质检结果执行真实筛选,输出符合条件的视频列表。", - "hyperparams": [ - { - "key": "DATE_START", - "value": "", - "description": "开始日期,支持 YYYY-MM-DD / YYYYMMDD", - "type": "string" - }, - { - "key": "DATE_END", - "value": "", - "description": "结束日期,支持 YYYY-MM-DD / YYYYMMDD", - "type": "string" - }, - { - "key": "PLACE", - "value": "", - "description": "采集地点", - "type": "string" - }, - { - "key": "TYPE", - "value": "", - "description": "采集场景类型", - "type": "string" - }, - { - "key": "MIN_CLARITY", - "value": "0.8", - "description": "清晰度最低阈值", - "type": "string" - }, - { - "key": "MAX_SHAKE", - "value": "0.1", - "description": "抖动度最高阈值", - "type": "string" - }, - { - "key": "MAX_HAND_OCCLUSION", - "value": "0.1", - "description": "无手率最高阈值", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/video_filter/operator.py", - "runtimeHandler": "builtin_dataset_filter" - }, - { - "id": "operator-builtin-video-slice", - "name": "视频切片", - "type": "builtin", - "category": "VIDEO", - "description": "将真实视频按固定时长切分为多个 mp4 片段。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "视频输入", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "clips", - "type": "json", - "description": "切片结果", - "transmission": "bus" - } - ], - "skillDefinition": "使用 ffmpeg/OpenCV 将真实视频切成固定长度片段,并输出 clip manifest。", - "hyperparams": [ - { - "key": "SEGMENT_DURATION_SECONDS", - "value": "300", - "description": "每个切片的秒数,显式覆盖时必须为正整数", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/video_slice/operator.py", - "runtimeHandler": "builtin_video_slicing" - }, - { - "id": "operator-builtin-video-annotate", - "name": "视频标注", - "type": "builtin", - "category": "VIDEO", - "description": "基于真实视频片段生成阶段划分、动作标签和对象提示。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "切片输入", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "annotations", - "type": "json", - "description": "标注结果", - "transmission": "bus" - } - ], - "skillDefinition": "读取切片视频并做真实帧间运动分析,输出 annotation JSON 和动作分段。", - "hyperparams": [ - { - "key": "FRAME_INTERVAL_SECONDS", - "value": "1.0", - "description": "采样帧间隔秒数,显式覆盖时必须为大于 0 的有限数字", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/video_annotate/operator.py", - "runtimeHandler": "builtin_video_annotation" - }, - { - "id": "operator-builtin-video-target-clip", - "name": "视频目标检测裁剪", - "type": "builtin", - "category": "VIDEO", - "description": "对真实视频执行目标出现区间检测,并自动剪出命中片段。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "视频清单或切片清单", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "targetClips", - "type": "json", - "description": "目标出现片段清单", - "transmission": "bus" - } - ], - "skillDefinition": "使用 OpenCV 对真实视频帧执行 motion / HOG person / OpenCV DNN 检测,按目标出现时间段剪辑 mp4 并输出 target clip manifest。", - "hyperparams": [ - { - "key": "DETECTION_MODE", - "value": "motion", - "description": "检测模式:motion、hog-person 或 opencv-dnn", - "type": "string" - }, - { - "key": "TARGET_LABELS", - "value": "motion,person,object", - "description": "目标标签,逗号分隔;motion 模式使用 motion,HOG 模式使用 person,DNN 模式使用模型类别名", - "type": "string" - }, - { - "key": "FRAME_INTERVAL_SECONDS", - "value": "0.5", - "description": "采样帧间隔秒数,显式覆盖时必须为大于 0 的有限数字", - "type": "string" - }, - { - "key": "MIN_CONFIDENCE", - "value": "0.2", - "description": "检测置信度阈值,范围 [0,1]", - "type": "string" - }, - { - "key": "MIN_MOTION_AREA_RATIO", - "value": "0.003", - "description": "motion 模式最小运动面积占比,范围 [0,1]", - "type": "string" - }, - { - "key": "MERGE_GAP_SECONDS", - "value": "0.5", - "description": "相邻命中段最大合并间隔秒数,显式覆盖时必须大于等于 0", - "type": "string" - }, - { - "key": "MIN_SEGMENT_SECONDS", - "value": "0.4", - "description": "输出片段最小时长秒数,显式覆盖时必须为大于 0 的有限数字", - "type": "string" - }, - { - "key": "PRE_ROLL_SECONDS", - "value": "0.1", - "description": "命中段前置缓冲秒数,显式覆盖时必须大于等于 0", - "type": "string" - }, - { - "key": "POST_ROLL_SECONDS", - "value": "0.2", - "description": "命中段后置缓冲秒数,显式覆盖时必须大于等于 0", - "type": "string" - }, - { - "key": "MAX_CLIPS", - "value": "20", - "description": "最多输出目标片段数量,显式覆盖时必须为正整数", - "type": "string" - }, - { - "key": "DNN_MODEL_PATH", - "value": "", - "description": "opencv-dnn 模式的真实模型文件路径,例如 ONNX;motion/HOG 模式无需配置", - "type": "string" - }, - { - "key": "DNN_CONFIG_PATH", - "value": "", - "description": "opencv-dnn 模式可选配置文件路径", - "type": "string" - }, - { - "key": "DNN_CLASS_NAMES", - "value": "", - "description": "opencv-dnn 模式可选类别名,逗号分隔;也可用 DNN_CLASS_NAMES_PATH 指向类别文件", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/video_target_clip/operator.py", - "runtimeHandler": "builtin_video_target_clip" - }, - { - "id": "operator-builtin-package", - "name": "成品打包", - "type": "builtin", - "category": "VIDEO", - "description": "将切片视频、质检结果、标注结果和配置文件组装成成品数据包。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "切片/标注/质检结果", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "packages", - "type": "json", - "description": "成品包清单", - "transmission": "bus" - } - ], - "skillDefinition": "按 clip 组织真实 package 目录,输出 packages manifest 和所有成品文件。", - "runtimeSource": "backend/operators/builtin/package/operator.py", - "runtimeHandler": "builtin_result_package" - }, - { - "id": "operator-builtin-audit-sample", - "name": "抽检采样", - "type": "builtin", - "category": "QA", - "description": "按质检和标注规则对成品数据执行真实抽检采样。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "成品包输入", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "samples", - "type": "json", - "description": "抽检样本清单", - "transmission": "bus" - } - ], - "skillDefinition": "根据 RULES_JSON 对成品数据包做规则筛选和稳定抽样,输出抽检任务清单。", - "hyperparams": [ - { - "key": "RULES_JSON", - "value": "[]", - "description": "抽检规则数组;显式规则必须包含 id/field/operator/value/samplePercent,operator 必须受支持,samplePercent 必须在 (0,100]", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/audit_sample/operator.py", - "runtimeHandler": "builtin_audit_sampling" - }, - { - "id": "operator-builtin-mongodb-write", - "name": "MongoDB 写入", - "type": "builtin", - "category": "DATA", - "description": "将上游 JSON 数据写入用户配置的业务 MongoDB,支持批量插入和按键 upsert,并允许显式覆盖连接。", - "inputs": [ - { - "id": "input", - "name": "input", - "type": "json", - "description": "待写入的 JSON 文档或 items 列表", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "result", - "name": "result", - "type": "json", - "description": "写入结果摘要", - "transmission": "bus" - } - ], - "skillDefinition": "将 JSON 文档写入业务 MongoDB 集合。", - "hyperparams": [ - { - "key": "MONGO_COLLECTION", - "value": "", - "description": "目标集合名称(禁止写系统保留集合)", - "type": "string" - }, - { - "key": "MONGO_MODE", - "value": "insert_many", - "description": "写入模式:insert_many 或 upsert", - "type": "string" - }, - { - "key": "MONGO_CONNECTION_ID", - "value": "", - "description": "业务 MongoDB 连接 ID;推荐优先配置", - "type": "string" - }, - { - "key": "MONGO_TARGET", - "value": "business", - "description": "兼容字段;默认 business。若显式传入 URI/DB,会优先使用显式连接", - "type": "string" - }, - { - "key": "UPSERT_KEY", - "value": "_id", - "description": "upsert 模式使用的唯一键", - "type": "string" - }, - { - "key": "DOCUMENT_PATH", - "value": "", - "description": "从输入 payload 中提取待写入文档的字段路径", - "type": "string" - }, - { - "key": "MONGO_DB", - "value": "", - "description": "可选数据库名;显式填写后会覆盖 business/platform 默认连接", - "type": "string" - }, - { - "key": "MONGO_URI", - "value": "", - "description": "可选 Mongo URI;显式填写后会覆盖 business/platform 默认连接", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/mongodb_write/operator.py", - "runtimeHandler": "builtin_mongodb_write" - }, - { - "id": "operator-builtin-hf-dataset-probe", - "name": "HF 数据集探测", - "type": "builtin", - "category": "DATASET", - "description": "调用 Hugging Face Dataset Viewer API,读取真实数据集的 splits、size、parquet 分片和样本行。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "可选上游数据集配置", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "profile", - "type": "json", - "description": "HF 数据集探测结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 Hugging Face Hub/LeRobot 等公开或私有数据集,真实调用 Dataset Viewer API 获取可用 split、样本行、parquet 分片和规模信息;不生成模拟数据。", - "hyperparams": [ - { - "key": "DATASET", - "value": "", - "description": "HF dataset repo id,例如 lerobot/pusht", - "type": "string" - }, - { - "key": "CONFIG", - "value": "", - "description": "可选 config;为空时从 splits 自动选择 train 或首个 split", - "type": "string" - }, - { - "key": "SPLIT", - "value": "", - "description": "可选 split;为空时优先选择 train", - "type": "string" - }, - { - "key": "ROW_OFFSET", - "value": "0", - "description": "样本行 offset,必须为非负整数", - "type": "string" - }, - { - "key": "ROW_LENGTH", - "value": "20", - "description": "样本行数量,范围 1-100", - "type": "string" - }, - { - "key": "INCLUDE_ROWS", - "value": "true", - "description": "是否调用 /rows 读取样本行", - "type": "string" - }, - { - "key": "INCLUDE_PARQUET", - "value": "true", - "description": "是否调用 /parquet 读取 parquet 分片", - "type": "string" - }, - { - "key": "INCLUDE_STATISTICS", - "value": "false", - "description": "是否调用 /statistics 读取列统计", - "type": "string" - }, - { - "key": "HF_TOKEN", - "value": "", - "description": "私有或 gated 数据集可填写 token;为空时不发送 Authorization", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/hf_dataset_probe/operator.py", - "runtimeHandler": "builtin_hf_dataset_probe" - }, - { - "id": "operator-builtin-parquet-table-summary", - "name": "Parquet 表摘要", - "type": "builtin", - "category": "DATASET", - "description": "读取真实 Parquet 文件或目录,汇总 schema、行数、row group、压缩编码和样本行,适配 Hugging Face/LeRobot 表格式数据入口。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "Parquet 文件、目录或上游 artifact", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "summary", - "type": "json", - "description": "Parquet 表结构和样本摘要", - "transmission": "bus" - } - ], - "skillDefinition": "面向 Hugging Face Dataset、LeRobot v3、轨迹元数据和 episode/action 表,真实读取 .parquet metadata 与样本行,校验必需列是否存在,不通过固定字段伪造输出。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "Parquet 文件或包含 .parquet 的目录", - "type": "string" - }, - { - "key": "REQUIRED_COLUMNS", - "value": "", - "description": "逗号分隔的必需列,例如 episode_index,frame_index,timestamp,action", - "type": "string" - }, - { - "key": "SAMPLE_COLUMNS", - "value": "", - "description": "逗号分隔的样本列;为空时读取首批全部列", - "type": "string" - }, - { - "key": "MAX_FILES", - "value": "50", - "description": "最多扫描 Parquet 文件数", - "type": "string" - }, - { - "key": "MAX_SAMPLE_ROWS", - "value": "5", - "description": "每个文件最多读取的样本行数;0 表示不读取样本行", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/parquet_table_summary/operator.py", - "runtimeHandler": "builtin_parquet_table_summary" - }, - { - "id": "operator-builtin-coco-annotation-validate", - "name": "COCO 标注校验", - "type": "builtin", - "category": "DATASET", - "description": "读取真实 COCO annotations JSON,校验 image/category 引用、bbox 边界、图片文件存在性和类别分布。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "COCO annotations JSON 或数据集根目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "COCO 标注校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向机器人感知、抓取检测、实例分割等 COCO 风格标注集,真实解析 images/annotations/categories,检查 bbox 是否越界、类别和图片引用是否存在,并统计 segmentation 覆盖。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "COCO annotation JSON 或数据集根目录", - "type": "string" - }, - { - "key": "IMAGE_ROOT", - "value": "", - "description": "可选图片根目录;为空时从 images/ 和 annotation 相邻目录推断", - "type": "string" - }, - { - "key": "MAX_ISSUES", - "value": "100", - "description": "每类问题最多保留样例数", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/coco_annotation_validate/operator.py", - "runtimeHandler": "builtin_coco_annotation_validate" - }, - { - "id": "operator-builtin-yolo-detection-validate", - "name": "YOLO 检测标注校验", - "type": "builtin", - "category": "DATASET", - "description": "读取真实 YOLO labels txt、images 目录和 data.yaml/classes,校验归一化 bbox、类别 ID、图片/标签配对和类别分布。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "YOLO 数据集根目录或 labels 文件", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "YOLO 检测标注校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向机器人目标检测、抓取区域识别和合成视觉数据训练,真实读取 YOLO txt 标注、data.yaml 类别和图片目录,检查 class id、归一化 bbox 范围、图片/标签成对关系;不接受空跑通过。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "YOLO 数据集根目录或 labels 文件", - "type": "string" - }, - { - "key": "CLASS_NAMES", - "value": "", - "description": "逗号分隔类别名;为空时读取 data.yaml/dataset.yaml 的 names", - "type": "string" - }, - { - "key": "IMAGE_ROOT", - "value": "", - "description": "可选图片根目录;为空时从 SOURCE_PATH 推断", - "type": "string" - }, - { - "key": "LABEL_ROOT", - "value": "", - "description": "可选标签根目录;为空时从 SOURCE_PATH 推断", - "type": "string" - }, - { - "key": "MAX_ISSUES", - "value": "100", - "description": "每类问题最多保留样例数", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/yolo_detection_validate/operator.py", - "runtimeHandler": "builtin_yolo_detection_validate" - }, - { - "id": "operator-builtin-dataset-split-manifest", - "name": "训练切分清单生成", - "type": "builtin", - "category": "DATASET", - "description": "基于真实 episode/demo 清单按稳定哈希生成 train/val/test 切分,并输出 JSON 与 CSV 清单。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "包含 episodes/demos/items 的上游摘要或真实清单", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "manifest", - "type": "json", - "description": "训练切分 JSON 清单", - "transmission": "bus" - }, - { - "id": "out-2", - "name": "csv", - "type": "file", - "description": "训练切分 CSV 清单", - "transmission": "bus" - } - ], - "skillDefinition": "面向 LeRobot、RLDS、HDF5/robomimic 等具身数据训练入口,读取真实 episode/demo metadata,用稳定哈希生成可复现的 train/val/test 清单;不按硬编码 ID 切分。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "可选 episode JSON/JSONL、LeRobot 根目录或包含 episodes.jsonl 的目录", - "type": "string" - }, - { - "key": "TRAIN_RATIO", - "value": "0.8", - "description": "训练集比例,必须为非负数字", - "type": "string" - }, - { - "key": "VAL_RATIO", - "value": "0.1", - "description": "验证集比例,必须为非负数字", - "type": "string" - }, - { - "key": "TEST_RATIO", - "value": "0.1", - "description": "测试集比例,必须为非负数字", - "type": "string" - }, - { - "key": "SEED", - "value": "droboticflow", - "description": "稳定哈希种子,同一数据和 seed 会生成同一切分", - "type": "string" - }, - { - "key": "MIN_STEPS", - "value": "1", - "description": "低于该 step 数的 episode/demo 会被跳过", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/dataset_split_manifest/operator.py", - "runtimeHandler": "builtin_dataset_split_manifest" - }, - { - "id": "operator-builtin-dataset-provenance-audit", - "name": "数据集来源许可审计", - "type": "builtin", - "category": "DATASET", - "description": "读取真实数据集目录中的 Dataset Card、LICENSE、CITATION 和 provenance metadata,输出来源许可审计结果。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "数据集目录或上游目录 artifact", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "audit", - "type": "json", - "description": "数据集来源许可审计结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 Hugging Face Dataset Card、GitHub 数据仓库和内部数据集发布前合规检查,真实读取 README front matter、LICENSE、CITATION、provenance.json 等文件,识别缺失 license/source/citation 的数据集。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "包含 README.md、LICENSE、CITATION 或 provenance.json 的真实数据集目录", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/dataset_provenance_audit/operator.py", - "runtimeHandler": "builtin_dataset_provenance_audit" - }, - { - "id": "operator-builtin-webdataset-shard-summary", - "name": "WebDataset 分片概览", - "type": "builtin", - "category": "DATASET", - "description": "读取真实 WebDataset tar/tar.gz 分片,统计样本键、扩展名、模态组合和不安全成员路径。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "WebDataset tar shard 文件或目录输入", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "summary", - "type": "json", - "description": "WebDataset shard 样本/模态概览", - "transmission": "bus" - } - ], - "skillDefinition": "面向 Hugging Face WebDataset、LAION 风格 tar shard 和内部大规模感知数据,真实读取 tar 成员并按同名 sample key 聚合图片、JSON、caption、tensor 等模态,不靠文件名硬编码结果。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": ".tar/.tar.gz 文件或包含 WebDataset 分片的目录", - "type": "string" - }, - { - "key": "MAX_SHARDS", - "value": "20", - "description": "最多扫描多少个 shard,必须为正整数", - "type": "string" - }, - { - "key": "MAX_PREVIEW_ENTRIES", - "value": "20", - "description": "每个 shard 最多返回多少个成员和样本预览", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/webdataset_shard_summary/operator.py", - "runtimeHandler": "builtin_webdataset_shard_summary" - }, - { - "id": "operator-builtin-agibot-world2026-validate", - "name": "AgiBotWorld2026 数据集校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 AgiBotWorld2026 meta/info、episode 元数据、data Parquet 和视频路径模板,校验 episode 长度、必需列、任务索引和视频文件可读性。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "AgiBotWorld2026 数据集根目录或上游目录 artifact", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "AgiBotWorld2026 parquet/video episode 结构校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 Hugging Face/ModelScope 上的 AgiBotWorld2026 大规模具身数据格式,真实读取 meta/info.json、meta/episodes.jsonl 或 meta/episodes/*.parquet、data/**/*.parquet,以及 video_path 模板指向的相机流文件,补齐训练前对 episode 行数、action/timestamp/frame_index 列、task 引用和视频缺失的验收。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "AgiBotWorld2026 数据集根目录或其中任意文件路径", - "type": "string" - }, - { - "key": "MAX_PARQUET_FILES", - "value": "100", - "description": "最多扫描多少个 episode/data Parquet 文件", - "type": "string" - }, - { - "key": "CHUNK_SIZE", - "value": "1000", - "description": "未显式提供 chunk_index 时用于 episode_index 推导 chunk 的大小", - "type": "string" - }, - { - "key": "REQUIRED_COLUMNS", - "value": "episode_index,frame_index,timestamp,action", - "description": "data Parquet 必须包含的列,逗号分隔", - "type": "string" - }, - { - "key": "DATA_PATH_TEMPLATE", - "value": "", - "description": "可选 data path 模板,例如 data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", - "type": "string" - }, - { - "key": "VIDEO_PATH_TEMPLATE", - "value": "", - "description": "可选 video path 模板,例如 videos/{video_key}/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.mp4", - "type": "string" - }, - { - "key": "VIDEO_KEYS", - "value": "", - "description": "可选视频 feature key 列表;为空时从 meta/info.json features 中 dtype=video 的字段推导", - "type": "string" - }, - { - "key": "CHECK_VIDEO_EXISTS", - "value": "true", - "description": "是否要求 video_path 模板对应的视频文件真实存在", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/agibot_world2026_validate/operator.py", - "runtimeHandler": "builtin_agibot_world2026_validate" - }, - { - "id": "operator-builtin-rlds-episode-summary", - "name": "RLDS 轨迹概览", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 RLDS/TFDS 导出的 episodes/steps JSON 或目录元数据,输出轨迹长度、动作维度、观测键和终止标志校验。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "RLDS JSON/JSONL 或目录输入", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "summary", - "type": "json", - "description": "RLDS 轨迹概览", - "transmission": "bus" - } - ], - "skillDefinition": "面向 Open X-Embodiment/RLDS 风格 trajectory 数据,读取真实 episode/step 结构并校验 is_first/is_last/is_terminal、reward、action、observation 等关键字段。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "RLDS JSON/JSONL 文件、导出目录或 TFDS 元数据目录", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/rlds_episode_summary/operator.py", - "runtimeHandler": "builtin_rlds_episode_summary" - }, - { - "id": "operator-builtin-bridgedata-tfds-manifest-validate", - "name": "BridgeData TFDS-RLDS清单校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 BridgeData/TFDS-RLDS 目录,校验 dataset_info/features 中的 image/action/language 特征声明,并逐条扫描 TFRecord shard framing 与 CRC。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "BridgeData/TFDS-RLDS 数据集目录、TFDS 导出目录或 TFRecord 文件", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "BridgeData TFDS-RLDS 清单校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 BridgeData V2、Open X-Embodiment 和 TFDS/RLDS 分片数据,真实读取 dataset_info.json、features.json 和 .tfrecord/.tfrec shard,校验 image/action/language_instruction 等训练关键字段声明、split record 数量、TFRecord length/header/payload CRC,避免只验证文件存在而漏掉损坏 shard 或缺少语言/动作特征。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "BridgeData/TFDS-RLDS 数据集目录或 TFRecord 文件", - "type": "string" - }, - { - "key": "MAX_FILES", - "value": "100", - "description": "最多扫描多少个 TFRecord shard", - "type": "string" - }, - { - "key": "MAX_RECORDS", - "value": "200000", - "description": "每个 shard 最多读取多少条 record", - "type": "string" - }, - { - "key": "PREVIEW_RECORDS", - "value": "3", - "description": "每个 shard 最多返回多少条 payload 预览", - "type": "string" - }, - { - "key": "IMAGE_FIELD_TOKENS", - "value": "image,observation.image,observation.images", - "description": "特征清单中用于判定图像观测字段的 token 列表", - "type": "string" - }, - { - "key": "ACTION_FIELD_TOKENS", - "value": "action", - "description": "特征清单中用于判定动作字段的 token 列表", - "type": "string" - }, - { - "key": "LANGUAGE_FIELD_TOKENS", - "value": "language_instruction,natural_language_instruction,instruction", - "description": "特征清单中用于判定语言指令字段的 token 列表", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/bridgedata_tfds_manifest_validate/operator.py", - "runtimeHandler": "builtin_bridgedata_tfds_manifest_validate" - }, - { - "id": "operator-builtin-droid-raw-episode-validate", - "name": "DROID Raw Episode校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 DROID raw episode 目录,校验 trajectory.h5、metadata_*.json 和 recordings/MP4 视频文件,并检查 action/state 数据集帧数一致性。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "DROID raw episode 目录或包含多个 episode 的根目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "DROID raw episode 文件结构和轨迹数据校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 DROID raw dataset 和 GitHub/HuggingFace 常见 DROID episode 导出,真实读取 trajectory.h5/trajectory_*.h5 HDF5 数据集、metadata_*.json 元数据和 recordings/MP4/*.mp4 视频,校验动作数据集、状态/关节/夹爪数据集、帧数一致性和视频文件存在性。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "DROID raw episode 目录或包含多个 episode 的根目录", - "type": "string" - }, - { - "key": "MAX_EPISODES", - "value": "100", - "description": "最多校验多少个 episode 目录,必须为正整数", - "type": "string" - }, - { - "key": "MAX_HDF5_ROWS", - "value": "200000", - "description": "每个 HDF5 dataset 最多抽样多少行用于有限值检查", - "type": "string" - }, - { - "key": "REQUIRE_ACTION", - "value": "true", - "description": "是否要求 HDF5 中存在 action/control 相关数据集", - "type": "boolean" - }, - { - "key": "REQUIRE_STATE", - "value": "true", - "description": "是否要求 HDF5 中存在 state/proprio/joint/gripper 相关数据集", - "type": "boolean" - }, - { - "key": "CHECK_VIDEO_EXISTS", - "value": "true", - "description": "是否要求 episode 中存在 recordings/MP4 下的 MP4 视频文件", - "type": "boolean" - } - ], - "runtimeSource": "backend/operators/builtin/droid_raw_episode_validate/operator.py", - "runtimeHandler": "builtin_droid_raw_episode_validate" - }, - { - "id": "operator-builtin-libero-hdf5-task-validate", - "name": "LIBERO HDF5任务校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 LIBERO HDF5 demonstration 文件,校验 /data/demo_*、actions、states、obs 图像/proprio 字段、语言任务字段和 BDDL 任务文件匹配关系。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "LIBERO .hdf5/.h5 文件或包含多个 task demo 的目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "LIBERO HDF5 task demonstration 结构校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 LIBERO、OpenVLA LIBERO fine-tuning 和 GitHub/HuggingFace 常见 LIBERO 数据集导出,真实读取 *_demo.hdf5 中 /data/demo_* 的 actions、states、obs/agentview_rgb、obs/eye_in_hand_rgb、obs/gripper_states、obs/joint_states、obs/ee_states、dones/rewards 和语言字段,并可核对同名 .bddl 任务定义文件。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "LIBERO .hdf5/.h5 文件或包含多个 task demo 的目录", - "type": "string" - }, - { - "key": "BDDL_PATH", - "value": "", - "description": "可选 BDDL 文件或目录;为空时会在 SOURCE_PATH 附近自动查找 .bddl", - "type": "string" - }, - { - "key": "MAX_FILES", - "value": "100", - "description": "最多校验多少个 HDF5 task 文件", - "type": "string" - }, - { - "key": "MAX_DEMOS", - "value": "1000", - "description": "每个文件最多读取多少个 demo group", - "type": "string" - }, - { - "key": "MAX_ROWS", - "value": "200000", - "description": "每个 dataset 最多抽样多少行用于有限值检查", - "type": "string" - }, - { - "key": "REQUIRED_OBS_DATASETS", - "value": "agentview_rgb,eye_in_hand_rgb,gripper_states,joint_states,ee_states", - "description": "必须存在的 obs 数据集路径,逗号分隔", - "type": "string" - }, - { - "key": "REQUIRE_BDDL", - "value": "false", - "description": "是否要求存在与 HDF5 task 同名的 .bddl 文件", - "type": "boolean" - }, - { - "key": "REQUIRE_STATES", - "value": "true", - "description": "是否要求每个 demo 包含 states 数据集", - "type": "boolean" - }, - { - "key": "REQUIRE_LANGUAGE", - "value": "false", - "description": "是否要求 HDF5 attrs 或 dataset 中存在 language/instruction/task 字段", - "type": "boolean" - }, - { - "key": "REQUIRE_TERMINAL_DONE", - "value": "false", - "description": "是否在存在 dones 数据集时要求最后一帧 done=true", - "type": "boolean" - } - ], - "runtimeSource": "backend/operators/builtin/libero_hdf5_task_validate/operator.py", - "runtimeHandler": "builtin_libero_hdf5_task_validate" - }, - { - "id": "operator-builtin-robocasa-hdf5-demonstration-validate", - "name": "RoboCasa HDF5示教校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 RoboCasa HDF5 demonstration 文件,校验 data/demo_*、actions、action_dict、states、obs 图像/proprio 字段、ep_meta 语言任务和 model_file 回放元数据。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "RoboCasa .hdf5/.h5 文件或包含多个 demonstration 文件的目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "RoboCasa HDF5 demonstration 结构校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 RoboCasa、robomimic/RoboSuite 派生家庭任务示教和 GitHub/HuggingFace 常见 RoboCasa 原始 HDF5 数据,真实读取 /data/demo_* 下 actions、action_dict、states、obs/next_obs,多相机图像、关节/夹爪/eef proprio 字段,并校验 demo.attrs.ep_meta、demo.attrs.model_file 和 data.attrs.env_args。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "RoboCasa .hdf5/.h5 文件或包含多个 demonstration 文件的目录", - "type": "string" - }, - { - "key": "MAX_FILES", - "value": "100", - "description": "最多校验多少个 HDF5 文件", - "type": "string" - }, - { - "key": "MAX_DEMOS", - "value": "1000", - "description": "每个文件最多读取多少个 demo group", - "type": "string" - }, - { - "key": "MAX_ROWS", - "value": "200000", - "description": "每个 dataset 最多抽样多少行用于有限值检查", - "type": "string" - }, - { - "key": "IMAGE_OBS_TOKENS", - "value": "image,rgb,depth,agentview,eye_in_hand", - "description": "用于识别图像观测字段的 token 列表", - "type": "string" - }, - { - "key": "PROPRIO_OBS_TOKENS", - "value": "state,joint,gripper,eef,ee_,proprio,qpos,qvel", - "description": "用于识别 proprio/机器人状态观测字段的 token 列表", - "type": "string" - }, - { - "key": "REQUIRE_ENV_ARGS", - "value": "true", - "description": "是否要求 /data attrs 中存在可解析的 env_args", - "type": "boolean" - }, - { - "key": "REQUIRE_MODEL_FILE", - "value": "true", - "description": "是否要求每个 demo 有 model_file 回放模型元数据", - "type": "boolean" - }, - { - "key": "REQUIRE_EP_META", - "value": "true", - "description": "是否要求每个 demo 有可解析的 ep_meta", - "type": "boolean" - }, - { - "key": "REQUIRE_ACTION_DICT", - "value": "true", - "description": "是否要求每个 demo 包含 action_dict 分量数据", - "type": "boolean" - }, - { - "key": "REQUIRE_STATES", - "value": "true", - "description": "是否要求每个 demo 包含 states 回放状态", - "type": "boolean" - }, - { - "key": "REQUIRE_LANGUAGE", - "value": "true", - "description": "是否要求 ep_meta 中存在 lang/language/task/instruction 字段", - "type": "boolean" - } - ], - "runtimeSource": "backend/operators/builtin/robocasa_hdf5_demonstration_validate/operator.py", - "runtimeHandler": "builtin_robocasa_hdf5_demonstration_validate" - }, - { - "id": "operator-builtin-furniturebench-pkl-demonstration-validate", - "name": "FurnitureBench PKL示教校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 FurnitureBench .pkl 示教文件,校验 observations、两路彩色图、robot_state、8 维 actions、rewards、skills 和 low/med/high 目录布局。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "FurnitureBench .pkl 文件或 low/med/high 数据集根目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "FurnitureBench PKL 示教结构校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 FurnitureBench 真实家具装配示教数据,读取 low/med/high//.pkl 或单个 .pkl 文件,校验 furniture、observations 列表、color_image1/color_image2、robot_state 中 ee_pos/ee_quat/关节/夹爪等状态、8-D actions、rewards 和 skills 的长度一致性、shape 与有限数值,适合在训练前发现损坏 pickle、缺失相机/状态字段和动作维度漂移。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "FurnitureBench .pkl 文件、家具目录或包含 low/med/high 的数据集根目录", - "type": "string" - }, - { - "key": "MAX_FILES", - "value": "100", - "description": "最多校验多少个 .pkl/.pickle 示教文件,必须为正整数", - "type": "string" - }, - { - "key": "EXPECTED_ACTION_DIM", - "value": "8", - "description": "期望 action 维度;FurnitureBench 官方格式为 8-D actions", - "type": "string" - }, - { - "key": "REQUIRED_IMAGE_KEYS", - "value": "color_image1,color_image2", - "description": "每帧 observation 必须包含的图像字段,默认 wrist/front 两路彩色相机", - "type": "string" - }, - { - "key": "REQUIRED_ROBOT_STATE_KEYS", - "value": "ee_pos,ee_quat,ee_pos_vel,ee_ori_vel,joint_positions,joint_velocities,joint_torques,gripper_width", - "description": "每帧 robot_state 必须包含的状态字段", - "type": "string" - }, - { - "key": "REQUIRE_SKILLS", - "value": "true", - "description": "是否要求 skills 列表存在且长度与 observations 一致", - "type": "boolean" - } - ], - "runtimeSource": "backend/operators/builtin/furniturebench_pkl_demonstration_validate/operator.py", - "runtimeHandler": "builtin_furniturebench_pkl_demonstration_validate" - }, - { - "id": "operator-builtin-roboturk-real-hdf5-validate", - "name": "RoboTurk Real HDF5校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取 RoboTurk Real Robot Dataset aligned HDF5,校验 data/user/demo 层级、视频路径属性、robot_observation ROS 观测和 user_control 控制序列。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "RoboTurk *_aligned_dataset.hdf5 文件或包含该文件的数据集目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "RoboTurk Real HDF5 结构与数值校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 RoboTurk Real Robot Dataset 的真实 aligned HDF5,读取 data//demo_* 层级,校验 total_demos/total_time、clock_time/session_id/demo_time/date/task、Kinect/USB 视频路径属性、robot_observation 中 eef_poses/eef_times/joint_states_arm/joint_states_arm_times/joint_states_gripper/joint_states_gripper_times,以及 user_control/user_control_times 的 shape、时间戳长度一致性和有限数值,适合在视频预测、模仿学习和真实机器人遥操作数据入库前发现结构漂移。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "RoboTurk *_aligned_dataset.hdf5 文件或数据集目录", - "type": "string" - }, - { - "key": "MAX_FILES", - "value": "20", - "description": "最多校验多少个 .hdf5/.h5 文件,必须为正整数", - "type": "string" - }, - { - "key": "MAX_DEMOS_PER_FILE", - "value": "50", - "description": "每个 HDF5 文件最多展开多少个 demo 明细,必须为正整数", - "type": "string" - }, - { - "key": "MAX_NUMERIC_SAMPLE_ROWS", - "value": "256", - "description": "每个数值 dataset 最多抽样多少行做 finite 检查", - "type": "string" - }, - { - "key": "REQUIRED_ROBOT_OBSERVATION_DATASETS", - "value": "eef_poses,eef_times,joint_states_arm,joint_states_arm_times,joint_states_gripper,joint_states_gripper_times", - "description": "robot_observation 中必须存在的数据集字段", - "type": "string" - }, - { - "key": "REQUIRED_USER_CONTROL_DATASETS", - "value": "user_control,user_control_times", - "description": "user_control 中必须存在的数据集字段", - "type": "string" - }, - { - "key": "REQUIRE_VIDEO_ATTRS", - "value": "true", - "description": "是否要求 top_depth_video_file、top_rgb_video_file、front_rgb_video_file 视频路径属性", - "type": "boolean" - }, - { - "key": "REQUIRE_FINITE_NUMERIC", - "value": "true", - "description": "是否抽样检查数值 dataset 不含 NaN/Inf", - "type": "boolean" - } - ], - "runtimeSource": "backend/operators/builtin/roboturk_real_hdf5_validate/operator.py", - "runtimeHandler": "builtin_roboturk_real_hdf5_validate" - }, - { - "id": "operator-builtin-roboset-hdf5-validate", - "name": "RoboSet HDF5轨迹校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取 RoboSet/RoboHive .h5/.hdf5 轨迹文件,校验 Trial*/data/derived/config 层级、actions、rewards、done/success、observation/state 和多相机 RGB 数据。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "RoboSet .h5/.hdf5 文件或包含多个轨迹文件的数据集目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "RoboSet HDF5 轨迹结构与数值校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 RoboSet/RoboHive 真实 HDF5 轨迹数据,读取 Trial0/Trial1/... 下的 data、derived、config,递归扫描 data 中 actions、rewards、done/success、observations/state/qpos/qvel、rgb/camera/image 多相机视觉数组,校验关键 group 存在、帧长一致性、RGB 视图数量、action 维度分布和有限数值,适合真实厨房遥操作、kinesthetic playback 与 RoboHive 数据入库前做格式漂移和损坏文件排查。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "RoboSet .h5/.hdf5 文件或数据集目录", - "type": "string" - }, - { - "key": "MAX_FILES", - "value": "20", - "description": "最多校验多少个 .h5/.hdf5 文件,必须为正整数", - "type": "string" - }, - { - "key": "MAX_TRIALS_PER_FILE", - "value": "50", - "description": "每个 HDF5 文件最多展开多少个 Trial 明细,必须为正整数", - "type": "string" - }, - { - "key": "MAX_NUMERIC_SAMPLE_ROWS", - "value": "256", - "description": "每个数值 dataset 最多抽样多少行做 finite 检查", - "type": "string" - }, - { - "key": "REQUIRED_TRIAL_KEYS", - "value": "data,derived,config", - "description": "每个 Trial group 必须存在的一级 key", - "type": "string" - }, - { - "key": "ACTION_KEY_CANDIDATES", - "value": "action,actions,u,ctrl,control", - "description": "识别 action 序列时使用的数据集候选名", - "type": "string" - }, - { - "key": "REWARD_KEY_CANDIDATES", - "value": "reward,rewards", - "description": "识别 reward 序列时使用的数据集候选名", - "type": "string" - }, - { - "key": "DONE_KEY_CANDIDATES", - "value": "done,dones,terminal,terminals,success,successes", - "description": "识别终止或成功标志序列时使用的数据集候选名", - "type": "string" - }, - { - "key": "OBSERVATION_KEY_CANDIDATES", - "value": "obs,observation,observations,state,states,qpos,qvel,proprio,proprioception", - "description": "识别 observation/state 序列时使用的数据集候选名", - "type": "string" - }, - { - "key": "RGB_KEY_TOKENS", - "value": "rgb,image,images,camera,cam", - "description": "识别多相机 RGB/image 数据集时使用的路径 token", - "type": "string" - }, - { - "key": "MIN_RGB_VIEWS", - "value": "1", - "description": "每个 Trial 至少需要识别到多少路 RGB/image 数据集;真实 kitchen 数据可设为 4", - "type": "string" - }, - { - "key": "REQUIRE_ACTIONS", - "value": "true", - "description": "是否要求每个 Trial 包含 action 序列", - "type": "boolean" - }, - { - "key": "REQUIRE_REWARDS", - "value": "true", - "description": "是否要求每个 Trial 包含 reward 序列", - "type": "boolean" - }, - { - "key": "REQUIRE_RGB", - "value": "true", - "description": "是否要求每个 Trial 包含 RGB/image 视觉序列", - "type": "boolean" - }, - { - "key": "REQUIRE_FINITE_NUMERIC", - "value": "true", - "description": "是否抽样检查数值 dataset 不含 NaN/Inf", - "type": "boolean" - } - ], - "runtimeSource": "backend/operators/builtin/roboset_hdf5_validate/operator.py", - "runtimeHandler": "builtin_roboset_hdf5_validate" - }, - { - "id": "operator-builtin-rh20t-scene-validate", - "name": "RH20T 场景结构校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 RH20T scene 目录,校验 metadata.json、cam_* 多相机媒体与 timestamps.npy、transformed 力/力矩 TCP 关节夹爪数组、高频数据和 calibration 标定文件。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "RH20T scene 目录或包含多个 scene 的根目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "RH20T scene 结构和关键数据文件校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 RH20T、Real Human 20T、真实遥操作多模态机器人数据,读取 scene folder 中 cam_* 相机目录、timestamps.npy、RGB/视频媒体、metadata.json、transformed/force_torque.npy、tcp.npy、joint.npy、gripper.npy、high_freq_data 和 calib 标定 npy 文件,检查目录结构、关键模态存在、时间戳单调性、数组 shape/finite 数值和 calibration 完整性。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "RH20T scene 目录或包含多个 scene 的根目录", - "type": "string" - }, - { - "key": "MAX_SCENES", - "value": "100", - "description": "最多校验多少个 scene 目录", - "type": "string" - }, - { - "key": "REQUIRED_TRANSFORMED_FILES", - "value": "force_torque.npy,tcp.npy,joint.npy,gripper.npy", - "description": "transformed 目录下必须存在的 .npy 文件,逗号分隔", - "type": "string" - }, - { - "key": "REQUIRE_METADATA", - "value": "true", - "description": "是否要求 scene 下存在可解析 metadata.json", - "type": "boolean" - }, - { - "key": "REQUIRE_CAMERAS", - "value": "true", - "description": "是否要求 scene 下存在 cam_* 相机目录", - "type": "boolean" - }, - { - "key": "REQUIRE_TRANSFORMED", - "value": "true", - "description": "是否要求 transformed 目录及关键力/位姿/关节/夹爪数组", - "type": "boolean" - }, - { - "key": "REQUIRE_CALIBRATION", - "value": "true", - "description": "是否要求 metadata calib 指向的 intrinsics/extrinsics/tcp 标定文件", - "type": "boolean" - } - ], - "runtimeSource": "backend/operators/builtin/rh20t_scene_validate/operator.py", - "runtimeHandler": "builtin_rh20t_scene_validate" - }, - { - "id": "operator-builtin-robomind-hdf5-quality-validate", - "name": "RoboMIND HDF5质量校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 RoboMIND trajectory.hdf5,校验 puppet/master joint_position、puppet/end_effector 卡死、非有限值、帧数和多模态字段。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "RoboMIND trajectory.hdf5 文件或包含 h5_* 轨迹目录的根目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "RoboMIND HDF5 轨迹质量校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 RoboMIND、Multi-embodiment Intelligence、真实遥操作 VLA 训练数据,读取 h5_franka_1rgb/h5_franka_3rgb/h5_ur_1rgb/h5_agilex_3rgb/h5_tienkung 等目录下的 trajectory.hdf5,解析 robot/task/split/episode 元信息,检查 puppet/end_effector、puppet/joint_position、master/joint_position、franka/* 候选字段的 shape、finite 数值、unique ratio 和卡死状态,并统计 RGB/image/camera 与 action/gripper/command 数据集。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "RoboMIND trajectory.hdf5 文件、data 目录或包含 benchmark/h5_* 的根目录", - "type": "string" - }, - { - "key": "MAX_FILES", - "value": "100", - "description": "最多校验多少个 .hdf5/.h5 轨迹文件,优先读取 trajectory.hdf5", - "type": "string" - }, - { - "key": "MAX_ROWS", - "value": "200000", - "description": "每个数值数组最多采样多少帧用于 range 和 unique ratio 计算", - "type": "string" - }, - { - "key": "EE_TOLERANCE", - "value": "0.001", - "description": "末端执行器数据 max(range_per_dim) 小于该阈值时判定卡死", - "type": "string" - }, - { - "key": "JOINT_TOLERANCE", - "value": "0.001", - "description": "关节数据 max(range_per_dim) 小于该阈值时判定卡死", - "type": "string" - }, - { - "key": "REQUIRE_JOINT_POSITION", - "value": "true", - "description": "是否要求存在 puppet/joint_position、puppet/joint_position_left 或 franka/joint_position", - "type": "boolean" - }, - { - "key": "REQUIRE_END_EFFECTOR", - "value": "false", - "description": "是否要求存在 puppet/end_effector、puppet/end_effector_left 或 franka/end_effector;默认仅在存在时检查质量", - "type": "boolean" - }, - { - "key": "REQUIRE_IMAGES", - "value": "false", - "description": "是否要求 HDF5 中存在 rgb/image/camera 相关观测数据集", - "type": "boolean" - } - ], - "runtimeSource": "backend/operators/builtin/robomind_hdf5_quality_validate/operator.py", - "runtimeHandler": "builtin_robomind_hdf5_quality_validate" - }, - { - "id": "operator-builtin-hdf5-demonstration-summary", - "name": "HDF5 示教数据概览", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 robomimic/DROID 风格 HDF5 示教文件,输出 demo 数、样本数、动作形状、观测键和 mask 信息。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "HDF5 文件或目录输入", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "summary", - "type": "json", - "description": "HDF5 示教数据概览", - "transmission": "bus" - } - ], - "skillDefinition": "面向 robomimic、DROID 等常见 HDF5 示教数据,真实读取 /data/demo_*、actions、states、obs、next_obs、mask 和 env_args 元数据;缺少 h5py 时直接失败。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": ".hdf5/.h5 文件或包含 HDF5 文件的目录", - "type": "string" - }, - { - "key": "MAX_DEMOS", - "value": "50", - "description": "最多返回多少个 demo 的摘要,必须为正整数", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/hdf5_demonstration_summary/operator.py", - "runtimeHandler": "builtin_hdf5_demonstration_summary" - }, - { - "id": "operator-builtin-robomimic-hdf5-validate", - "name": "robomimic HDF5 示教校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 robomimic HDF5 示教文件,校验 /data/demo_* 下 actions、rewards、dones、obs、next_obs、states、mask 与 total 样本数一致性。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "robomimic .hdf5/.h5 文件或目录输入", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "robomimic HDF5 示教结构校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 robomimic、D4RL/robosuite 派生示教数据和 GitHub 常见机器人模仿学习 HDF5 数据,真实读取 /data/demo_* group,校验 actions/rewards/dones 与 obs/next_obs 帧数、num_samples、data.attrs.total 和 mask/train/valid 引用关系。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": ".hdf5/.h5 文件或包含 robomimic HDF5 文件的目录", - "type": "string" - }, - { - "key": "REQUIRE_REWARDS", - "value": "true", - "description": "是否要求每个 demo 包含 rewards 数据集", - "type": "string" - }, - { - "key": "REQUIRE_DONES", - "value": "true", - "description": "是否要求每个 demo 包含 dones 数据集", - "type": "string" - }, - { - "key": "REQUIRE_NEXT_OBS", - "value": "true", - "description": "是否要求每个 demo 包含 next_obs group", - "type": "string" - }, - { - "key": "CHECK_ACTION_RANGE", - "value": "false", - "description": "是否检查 action 绝对值不超过 MAX_ABS_ACTION;默认关闭以兼容未归一化控制量", - "type": "string" - }, - { - "key": "MAX_ABS_ACTION", - "value": "1.0", - "description": "开启 CHECK_ACTION_RANGE 后允许的最大 action 绝对值", - "type": "string" - }, - { - "key": "MAX_DEMOS", - "value": "200", - "description": "每个文件最多校验多少个 demo,必须为正整数", - "type": "string" - }, - { - "key": "MAX_FILES", - "value": "20", - "description": "最多校验多少个 HDF5 文件,必须为正整数", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/robomimic_hdf5_validate/operator.py", - "runtimeHandler": "builtin_robomimic_hdf5_validate" - }, - { - "id": "operator-builtin-aloha-hdf5-validate", - "name": "ALOHA HDF5 示教校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 ACT/ALOHA 风格 HDF5 示教文件,校验 action、observations/qpos、observations/qvel 和多相机图像帧数一致性。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "ALOHA/ACT .hdf5/.h5 文件或目录输入", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "ALOHA HDF5 示教结构校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 ALOHA、Mobile ALOHA、ACT 等双臂示教数据,真实读取 /action、/observations/qpos、/observations/qvel 和 /observations/images/ HDF5 数据集,校验缺失数据、帧数漂移、相机数量和 action/qpos 维度。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": ".hdf5/.h5 文件或包含 ALOHA HDF5 文件的目录", - "type": "string" - }, - { - "key": "ACTION_PATH", - "value": "/action", - "description": "HDF5 中动作数组路径", - "type": "string" - }, - { - "key": "QPOS_PATH", - "value": "/observations/qpos", - "description": "HDF5 中关节位置数组路径", - "type": "string" - }, - { - "key": "QVEL_PATH", - "value": "/observations/qvel", - "description": "HDF5 中关节速度数组路径", - "type": "string" - }, - { - "key": "IMAGE_GROUP_PATH", - "value": "/observations/images", - "description": "HDF5 中多相机图像 group 路径", - "type": "string" - }, - { - "key": "REQUIRE_QVEL", - "value": "false", - "description": "是否要求 qvel 数据集必须存在", - "type": "string" - }, - { - "key": "MAX_FILES", - "value": "20", - "description": "最多校验多少个 HDF5 文件", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/aloha_hdf5_validate/operator.py", - "runtimeHandler": "builtin_aloha_hdf5_validate" - }, - { - "id": "operator-builtin-umi-zarr-dataset-validate", - "name": "UMI Zarr数据集校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 UMI dataset.zarr 或 dataset.zarr.zip,校验 data 数组、meta/episode_ends、首维时间步一致性、关键末端/夹爪/相机字段和相机 chunk 配置。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "UMI dataset.zarr 目录、dataset.zarr.zip 文件或其上级目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "UMI Zarr 数据集结构和关键训练数组校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 UMI、Universal Manipulation Interface、Diffusion Policy Zarr 训练数据和社区 UMI-like 数据,真实打开 dataset.zarr 或 dataset.zarr.zip,读取 root/data 下 camera*_rgb、robot0_eef_pos、robot0_eef_rot_axis_angle、robot0_gripper_width 等数组,以及 root/meta/episode_ends,校验 episode 边界、所有 data 数组首维时间步一致、关键字段存在、数值 finite 和相机时间维 chunk 是否适合随机读取训练。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "UMI dataset.zarr 目录、dataset.zarr.zip 文件或包含它们的目录", - "type": "string" - }, - { - "key": "MAX_ARRAYS", - "value": "100", - "description": "最多校验并返回多少个 data array 摘要,必须为正整数", - "type": "string" - }, - { - "key": "MAX_SAMPLE_ROWS", - "value": "256", - "description": "非图像数值数组最多抽样多少帧用于 finite/min/max 检查", - "type": "string" - }, - { - "key": "REQUIRED_DATA_KEYS", - "value": "robot0_eef_pos,robot0_eef_rot_axis_angle,robot0_gripper_width", - "description": "逗号分隔的 UMI data 关键数组名;为空时使用默认末端位姿和夹爪宽度字段", - "type": "string" - }, - { - "key": "REQUIRE_CAMERA_RGB", - "value": "true", - "description": "是否要求存在 camera*_rgb 形态的 RGB 图像数组", - "type": "boolean" - }, - { - "key": "CHECK_CHUNKING", - "value": "true", - "description": "是否检查相机数组时间维 chunk 不超过 MAX_CAMERA_CHUNK_FRAMES", - "type": "boolean" - }, - { - "key": "MAX_CAMERA_CHUNK_FRAMES", - "value": "4", - "description": "相机 RGB 数组允许的最大时间维 chunk 帧数;UMI 官方示例为 1", - "type": "string" - }, - { - "key": "MAX_EPISODE_ENDS", - "value": "100", - "description": "产物中最多返回多少个 episode_ends 值;校验仍读取完整数组", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/umi_zarr_dataset_validate/operator.py", - "runtimeHandler": "builtin_umi_zarr_dataset_validate" - }, - { - "id": "operator-builtin-zarr-trajectory-summary", - "name": "Zarr 轨迹数组概览", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 Zarr group/array store,输出 actions、states、episode_ends 等数组形状、chunk 和 dtype 概览。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "Zarr 目录或上游目录 artifact", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "summary", - "type": "json", - "description": "Zarr 轨迹数组概览", - "transmission": "bus" - } - ], - "skillDefinition": "面向 Diffusion Policy、ManiSkill、内部机器人数组轨迹等 Zarr 数据,真实读取 .zgroup/.zarray/.zattrs 或 zarr.json 元数据,汇总数组 shape、chunk、dtype、episode_count 和 step_count。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "Zarr group/array store 根目录", - "type": "string" - }, - { - "key": "MAX_ARRAYS", - "value": "100", - "description": "最多返回多少个 array 摘要,必须为正整数", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/zarr_trajectory_summary/operator.py", - "runtimeHandler": "builtin_zarr_trajectory_summary" - }, - { - "id": "operator-builtin-rlbench-dataset-validate", - "name": "RLBench 数据集校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 RLBench task/variation/episode 目录,校验 low_dim_obs.pkl、variation_descriptions.pkl 和多相机 RGB/depth/mask 帧数量一致性。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "RLBench task、variation 或 episode 目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "RLBench 数据集校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 RLBench、RVT、PerAct 等常见多任务仿真演示数据,真实读取 episode 下 low_dim_obs.pkl 和 variation_descriptions.pkl,并扫描 front/left_shoulder/right_shoulder/wrist 等相机 RGB、depth、mask 目录,检查低维观测步数、语言描述和图像帧数是否对齐。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "RLBench episode/task/variation 目录或其中任意文件路径", - "type": "string" - }, - { - "key": "CAMERA_DIRS", - "value": "front_rgb,left_shoulder_rgb,right_shoulder_rgb,wrist_rgb,overhead_rgb,front_depth,left_shoulder_depth,right_shoulder_depth,wrist_depth,front_mask,left_shoulder_mask,right_shoulder_mask,wrist_mask", - "description": "需要扫描的相机目录名,逗号分隔", - "type": "string" - }, - { - "key": "MAX_EPISODES", - "value": "200", - "description": "最多校验多少个 episode 目录,必须为正整数", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/rlbench_dataset_validate/operator.py", - "runtimeHandler": "builtin_rlbench_dataset_validate" - }, - { - "id": "operator-builtin-maniskill-trajectory-validate", - "name": "ManiSkill 轨迹校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 ManiSkill .h5/.hdf5 轨迹和同名 JSON metadata,校验 trajectory group、actions、rewards/terminated/env_states 长度和 metadata elapsed_steps。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "ManiSkill HDF5 轨迹文件或目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "ManiSkill 轨迹校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 ManiSkill2/3、SAPIEN 仿真演示和 GitHub 常见 HDF5 轨迹数据,真实读取 .h5/.hdf5 里的 traj_* group、actions、rewards、terminated、env_states 等 dataset,并读取同名 JSON metadata 的 env_info/episodes,检查动作步数、轨迹数量和 metadata elapsed_steps 是否一致。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "ManiSkill .h5/.hdf5 轨迹文件或包含轨迹文件的目录", - "type": "string" - }, - { - "key": "MAX_FILES", - "value": "50", - "description": "最多读取多少个 HDF5 文件", - "type": "string" - }, - { - "key": "MAX_ACTION_ROWS", - "value": "200000", - "description": "每个 actions dataset 最多扫描多少行用于有限值检查", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/maniskill_trajectory_validate/operator.py", - "runtimeHandler": "builtin_maniskill_trajectory_validate" - }, - { - "id": "operator-builtin-calvin-episode-validate", - "name": "CALVIN Episode 校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 CALVIN episode_*.npz 和 lang_annotations/auto_lang_ann.npy,校验 actions、rel_actions、robot_obs、scene_obs、rgb_static、rgb_gripper 等数组。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "CALVIN 数据集目录或 episode .npz 文件", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "CALVIN Episode 校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 CALVIN、语言条件长程操作和多相机桌面机器人数据,真实读取 episode_*.npz 中 actions、rel_actions、robot_obs、scene_obs、rgb_static、rgb_gripper 等数组,检查必需数组存在、shape/dtype、有限数值和图像数组维度,并解析 lang_annotations/auto_lang_ann.npy 的语言标注数量。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "CALVIN 数据集目录或 episode .npz 文件", - "type": "string" - }, - { - "key": "REQUIRED_ARRAYS", - "value": "actions,rel_actions,robot_obs,scene_obs,rgb_static,rgb_gripper", - "description": "必需 NPZ 数组名称,逗号分隔", - "type": "string" - }, - { - "key": "MAX_FILES", - "value": "500", - "description": "最多读取多少个 episode .npz 文件", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/calvin_episode_validate/operator.py", - "runtimeHandler": "builtin_calvin_episode_validate" - }, - { - "id": "operator-builtin-rtx-action-schema-validate", - "name": "RT-X动作空间Schema校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 CSV/JSON/JSONL/Parquet/HDF5 动作数据,按 RT-X/Open X 常见 7 维动作空间校验 action 维度、gripper 范围、有限值和动作幅度。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "可选上游数据;通常使用 SOURCE_PATH 指向真实动作文件或目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "RT-X 动作空间校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 Open X-Embodiment / RT-X / BridgeData 等统一动作空间训练前验收,检查动作向量是否符合配置维度,默认 7 维末位 gripper,并约束 gripper 与 delta 幅度。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "真实 CSV/JSON/JSONL/Parquet/HDF5 动作文件或目录", - "type": "string" - }, - { - "key": "ACTION_FIELDS", - "value": "action,actions,world_vector,rotation_delta,gripper_closedness_action", - "description": "候选动作字段,支持点路径,逗号分隔", - "type": "string" - }, - { - "key": "EXPECTED_ACTION_DIM", - "value": "7", - "description": "期望动作向量维度,RT-X 常见为 7", - "type": "number" - }, - { - "key": "GRIPPER_INDEX", - "value": "6", - "description": "gripper 维度索引,默认第 7 维", - "type": "number" - }, - { - "key": "GRIPPER_MIN", - "value": "-1", - "description": "gripper 允许最小值", - "type": "number" - }, - { - "key": "GRIPPER_MAX", - "value": "1", - "description": "gripper 允许最大值", - "type": "number" - }, - { - "key": "MAX_ABS_DELTA", - "value": "10", - "description": "动作每个维度允许的最大绝对值", - "type": "number" - }, - { - "key": "REQUIRE_GRIPPER", - "value": "true", - "description": "是否要求 gripper 维度存在且在范围内", - "type": "boolean" - }, - { - "key": "MAX_FILES", - "value": "50", - "description": "最多扫描的数据文件数", - "type": "number" - }, - { - "key": "MAX_ROWS", - "value": "200000", - "description": "最多扫描的动作行数", - "type": "number" - } - ], - "runtimeSource": "backend/operators/builtin/rtx_action_schema_validate/operator.py", - "runtimeHandler": "builtin_rtx_action_schema_validate" - }, - { - "id": "operator-builtin-embodiment-metadata-validate", - "name": "多机器人Embodiment元信息校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 CSV/JSON/JSONL/Parquet 元信息或 manifest,校验 robot/embodiment、task、camera、action_dim 等字段并统计多机器人覆盖。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "可选上游数据;通常使用 SOURCE_PATH 指向真实元信息或 manifest", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "Embodiment 元信息校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 RoboMIND、Open X-Embodiment 等多机器人数据集,统一验收机器人类型、任务、相机命名和动作维度元信息,避免混合数据训练前 metadata 不可追溯。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "真实元信息或 manifest 文件/目录", - "type": "string" - }, - { - "key": "REQUIRED_FIELDS", - "value": "robot_type,task,action_dim", - "description": "必须存在的语义字段,可包含 robot_type、task、camera、action_dim", - "type": "string" - }, - { - "key": "MIN_ROBOT_TYPES", - "value": "1", - "description": "期望至少覆盖多少种机器人类型", - "type": "number" - }, - { - "key": "ROBOT_FIELDS", - "value": "robot_type,robotType,robot,embodiment,embodiment_name,platform", - "description": "候选机器人/embodiment 字段,逗号分隔", - "type": "string" - }, - { - "key": "TASK_FIELDS", - "value": "task,task_name,taskName,language_instruction,instruction", - "description": "候选任务/语言指令字段,逗号分隔", - "type": "string" - }, - { - "key": "CAMERA_FIELDS", - "value": "camera_names,cameras,camera,observation.images,image_modalities", - "description": "候选相机字段,逗号分隔", - "type": "string" - }, - { - "key": "ACTION_DIM_FIELDS", - "value": "action_dim,actionDimension,action_space_dim,action_shape", - "description": "候选动作维度字段,逗号分隔", - "type": "string" - }, - { - "key": "MAX_FILES", - "value": "50", - "description": "最多扫描的元信息文件数", - "type": "number" - }, - { - "key": "MAX_ROWS", - "value": "100000", - "description": "最多扫描的元信息行数", - "type": "number" - } - ], - "runtimeSource": "backend/operators/builtin/embodiment_metadata_validate/operator.py", - "runtimeHandler": "builtin_embodiment_metadata_validate" - }, - { - "id": "operator-builtin-force-torque-sensor-validate", - "name": "力扭矩传感器数据校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 CSV/JSON/JSONL/Parquet/HDF5 力/扭矩/wrench 数据,校验 3D/6D 维度、有限值、阈值范围和时间戳单调性。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "可选上游数据;通常使用 SOURCE_PATH 指向真实力/扭矩传感器文件或目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "力/扭矩传感器校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向接触丰富操作、失败检测、装配和遥操作回放数据,真实读取 wrench/force/torque 序列,检查传感器数据是否可训练、可同步、未饱和。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "真实 CSV/JSON/JSONL/Parquet/HDF5 力/扭矩传感器文件或目录", - "type": "string" - }, - { - "key": "EXPECTED_DIM", - "value": "6", - "description": "期望力/扭矩向量维度,常见 wrench 为 6", - "type": "number" - }, - { - "key": "MAX_FORCE_ABS", - "value": "10000", - "description": "前三个 force 维度的最大绝对值阈值", - "type": "number" - }, - { - "key": "MAX_TORQUE_ABS", - "value": "10000", - "description": "torque 维度的最大绝对值阈值", - "type": "number" - }, - { - "key": "CHECK_TIMESTAMP_MONOTONIC", - "value": "true", - "description": "是否校验时间戳单调不回退", - "type": "boolean" - }, - { - "key": "WRENCH_FIELDS", - "value": "wrench,force_torque,forceTorque,ft,sensor.wrench", - "description": "候选 wrench 字段,逗号分隔", - "type": "string" - }, - { - "key": "FORCE_FIELDS", - "value": "force,forces,sensor.force", - "description": "候选 force 字段,逗号分隔", - "type": "string" - }, - { - "key": "TORQUE_FIELDS", - "value": "torque,torques,sensor.torque", - "description": "候选 torque 字段,逗号分隔", - "type": "string" - }, - { - "key": "TIMESTAMP_FIELDS", - "value": "timestamp,time,t,frame_timestamp,sensor.timestamp", - "description": "候选时间戳字段,逗号分隔", - "type": "string" - }, - { - "key": "MAX_FILES", - "value": "50", - "description": "最多扫描的数据文件数", - "type": "number" - }, - { - "key": "MAX_ROWS", - "value": "200000", - "description": "最多扫描的数据行数", - "type": "number" - } - ], - "runtimeSource": "backend/operators/builtin/force_torque_sensor_validate/operator.py", - "runtimeHandler": "builtin_force_torque_sensor_validate" - }, - { - "id": "operator-builtin-trajectory-integrity-validate", - "name": "轨迹多模态完整性校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 episode/step JSON 或 JSONL,校验动作、状态、图像引用等多模态字段完整性和维度一致性。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "包含 episodes/steps 的轨迹 JSON/JSONL 或上游 artifact", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "轨迹多模态完整性校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 Open X-Embodiment、BridgeData、DROID、LeRobot 等具身轨迹数据,逐 step 校验 action、observation、image/state 等字段和本地文件引用,输出可审计缺失项。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "episode JSON/JSONL 文件或包含 episodes.jsonl 的目录", - "type": "string" - }, - { - "key": "REQUIRED_FIELDS", - "value": "action,observation.image,observation.state", - "description": "逗号分隔的必需字段路径,支持 observation.image 等点路径", - "type": "string" - }, - { - "key": "CHECK_FILE_EXISTS", - "value": "true", - "description": "是否校验 observation.image 等相对路径在数据目录中真实存在", - "type": "string" - }, - { - "key": "REQUIRE_CONSISTENT_ACTION_DIM", - "value": "true", - "description": "是否要求所有 step 的 action 维度一致", - "type": "string" - }, - { - "key": "REQUIRE_CONSISTENT_STATE_DIM", - "value": "true", - "description": "是否要求所有 step 的 observation.state 维度一致", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/trajectory_integrity_validate/operator.py", - "runtimeHandler": "builtin_trajectory_integrity_validate" - }, - { - "id": "operator-builtin-action-state-quality-validate", - "name": "动作状态数值质量校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 JSON/JSONL/Parquet 轨迹表,校验 action/state 缺失、维度一致、非有限值和异常幅值。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "包含 action/state 的轨迹 JSON、JSONL、Parquet 文件或目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "动作状态数值质量校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 LeRobot、DROID、Open X-Embodiment/RLDS 导出的训练表,真实读取 action、observation.state、joint_positions 等字段,检查缺失、非数字、NaN/Inf、维度漂移和异常大数值,并输出每个维度的 min/max/mean。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "轨迹 JSON/JSONL/Parquet 文件或目录", - "type": "string" - }, - { - "key": "ACTION_FIELDS", - "value": "action,actions,control,command", - "description": "候选动作字段,逗号分隔;支持 observation.state 这样的点路径", - "type": "string" - }, - { - "key": "STATE_FIELDS", - "value": "observation.state,state,states,observation.joint_positions,joint_positions,proprio", - "description": "候选状态/本体感知字段,逗号分隔", - "type": "string" - }, - { - "key": "MAX_ROWS", - "value": "10000", - "description": "最多读取多少行真实数据", - "type": "string" - }, - { - "key": "MAX_FILES", - "value": "50", - "description": "最多读取多少个输入文件", - "type": "string" - }, - { - "key": "MAX_ABS_VALUE", - "value": "1000000", - "description": "action/state 元素允许的最大绝对值,超过会记为异常幅值", - "type": "string" - }, - { - "key": "LOW_VARIANCE_EPSILON", - "value": "0.000000001", - "description": "维度低方差提示阈值;仅作为 warning,不直接判失败", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/action_state_quality_validate/operator.py", - "runtimeHandler": "builtin_action_state_quality_validate" - }, - { - "id": "operator-builtin-episode-boundary-validate", - "name": "Episode 边界校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 episode JSON/JSONL/Parquet,按 episode 分组校验 frame_index 连续性和终止标志位置。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "包含 episode/frame/done 字段的轨迹文件或目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "Episode 边界校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 RLDS、LeRobot、DROID、ALOHA 等 episode 化轨迹,真实读取 episode_index、frame_index、is_last/done/terminated/success 等字段,检查每个 episode 是否从 0 开始、frame 是否连续递增、终止标志是否只出现在末尾。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "episode JSON/JSONL/Parquet 文件或目录", - "type": "string" - }, - { - "key": "EPISODE_FIELDS", - "value": "episode_index,episode_id,episodeId,trajectory_id,trajectoryId", - "description": "候选 episode ID 字段,逗号分隔", - "type": "string" - }, - { - "key": "FRAME_FIELDS", - "value": "frame_index,step_index,timestep,time_step", - "description": "候选帧/步索引字段,逗号分隔", - "type": "string" - }, - { - "key": "TERMINAL_FIELDS", - "value": "is_last,is_terminal,done,terminated,success", - "description": "候选终止标志字段,逗号分隔", - "type": "string" - }, - { - "key": "REQUIRE_ZERO_FRAME_START", - "value": "true", - "description": "是否要求每个 episode 的第一帧索引为 0", - "type": "string" - }, - { - "key": "REQUIRE_TERMINAL_FLAG", - "value": "true", - "description": "是否要求每个 episode 最后一帧有终止标志", - "type": "string" - }, - { - "key": "MAX_ROWS", - "value": "100000", - "description": "最多读取多少行真实数据", - "type": "string" - }, - { - "key": "MAX_FILES", - "value": "50", - "description": "最多读取多少个输入文件", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/episode_boundary_validate/operator.py", - "runtimeHandler": "builtin_episode_boundary_validate" - }, - { - "id": "operator-builtin-rgbd-frame-pair-validate", - "name": "RGB-D 帧对齐校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 RGB 与 depth 图像目录,按帧名配对,校验尺寸一致性、深度有效像素、深度范围和相机内参。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "RGB-D 数据集目录或上游目录 artifact", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "RGB-D 帧对齐校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 ManiSkill、真实机器人 RGB-D 采集、仿真深度图和 VLA 视觉观测入库,真实读取 rgb/color/images 与 depth/depths PNG/TIFF/NPY,校验同名帧是否成对、图像尺寸是否一致、depth 是否有有效像素,并解析 fx/fy/cx/cy 内参。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "RGB-D 数据集根目录", - "type": "string" - }, - { - "key": "RGB_DIR", - "value": "", - "description": "可选 RGB 目录;为空时自动从 rgb/color/images 推断", - "type": "string" - }, - { - "key": "DEPTH_DIR", - "value": "", - "description": "可选 depth 目录;为空时自动从 depth/depths/images/depth 推断", - "type": "string" - }, - { - "key": "INTRINSICS_PATH", - "value": "", - "description": "可选 intrinsics JSON/YAML;为空时自动查找 camera_intrinsics.json/intrinsics.json/camera.yaml", - "type": "string" - }, - { - "key": "DEPTH_SCALE", - "value": "1000", - "description": "depth 原始单位到米的缩放分母,例如 uint16 毫米深度使用 1000", - "type": "string" - }, - { - "key": "MAX_PAIRS", - "value": "200", - "description": "最多校验多少对 RGB-D 帧", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/rgbd_frame_pair_validate/operator.py", - "runtimeHandler": "builtin_rgbd_frame_pair_validate" - }, - { - "id": "operator-builtin-segmentation-mask-validate", - "name": "分割 Mask 校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实语义/实例分割 mask 与可选原图,校验 mask/图像配对、尺寸一致性、类别 ID 范围和像素分布。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "分割 mask 数据集目录或上游目录 artifact", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "分割 mask 校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向仿真语义分割、机器人场景理解、抓取区域 mask 和实例分割训练数据,真实读取 masks/segmentation/labels/semantic PNG/TIFF/NPY,统计 class id 像素分布,校验图像/Mask 成对关系与尺寸一致性。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "分割数据集根目录", - "type": "string" - }, - { - "key": "IMAGE_DIR", - "value": "", - "description": "可选原图目录;为空时从 images/rgb/color 推断", - "type": "string" - }, - { - "key": "MASK_DIR", - "value": "", - "description": "可选 mask 目录;为空时从 masks/segmentation/labels/semantic 推断", - "type": "string" - }, - { - "key": "CLASS_IDS", - "value": "", - "description": "可选允许的类别 ID,逗号分隔;为空时只统计不限制", - "type": "string" - }, - { - "key": "MAX_PAIRS", - "value": "200", - "description": "最多校验多少个 mask 或图像/mask 对", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/segmentation_mask_validate/operator.py", - "runtimeHandler": "builtin_segmentation_mask_validate" - }, - { - "id": "operator-builtin-language-instruction-validate", - "name": "语言指令校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 JSONL/CSV/Parquet/JSON 元数据中的 language_instruction/instruction/task 等字段,统计覆盖率、去重和任务分布。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "包含语言指令的 episode/task 元数据文件或目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "语言指令覆盖与质量校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 Open X-Embodiment、RLDS、LeRobot、VLA 训练集和人工采集任务元数据,真实读取 episodes/tasks JSONL、CSV、Parquet 或 JSON,检查 language_instruction/instruction/task/command/prompt 字段覆盖率、空指令、重复指令和 episode/task 分组数量。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "元数据文件或包含 episodes/tasks 的目录", - "type": "string" - }, - { - "key": "TEXT_FIELDS", - "value": "language_instruction,instruction,task,command,prompt", - "description": "逗号分隔的候选语言字段名", - "type": "string" - }, - { - "key": "GROUP_FIELDS", - "value": "episode_index,episode_id,trajectory_id,task_index,task_id", - "description": "逗号分隔的 episode/task 分组字段名", - "type": "string" - }, - { - "key": "MAX_ROWS", - "value": "10000", - "description": "最多读取多少行元数据", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/language_instruction_validate/operator.py", - "runtimeHandler": "builtin_language_instruction_validate" - }, - { - "id": "operator-builtin-vla-sample-manifest-validate", - "name": "VLA 样本清单校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 VLA/OpenVLA/DROID 风格 JSON、JSONL、CSV 或 Parquet 样本清单,校验图像引用、语言指令和动作向量完整性。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "包含图像、语言指令和动作字段的 VLA 样本清单文件或目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "VLA 样本清单校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 OpenVLA、DROID、Bridge/Open X-Embodiment 派生 VLA 训练清单,真实读取 image/image_path/observation.image、language_instruction/instruction/conversations 和 action/control 字段,校验本地图像引用存在、语言指令覆盖、动作维度一致和完整样本数量。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "VLA 样本 JSON/JSONL/CSV/Parquet 文件或目录", - "type": "string" - }, - { - "key": "IMAGE_FIELDS", - "value": "image,image_path,images,observation.image,observation.images,observation.exterior_image_1_left,observation.exterior_image_2_left,observation.wrist_image_left,exterior_image_1_left,exterior_image_2_left,wrist_image_left", - "description": "逗号分隔的候选图像引用字段,支持点路径", - "type": "string" - }, - { - "key": "TEXT_FIELDS", - "value": "language_instruction,instruction,task,prompt,goal,conversations", - "description": "逗号分隔的候选语言字段,支持 conversations 列表", - "type": "string" - }, - { - "key": "ACTION_FIELDS", - "value": "action,actions,control,controls,target_action", - "description": "逗号分隔的候选动作字段", - "type": "string" - }, - { - "key": "CHECK_IMAGE_EXISTS", - "value": "true", - "description": "是否校验本地图像路径真实存在;远程 URI 只计数不下载", - "type": "string" - }, - { - "key": "REQUIRE_CONSISTENT_ACTION_DIM", - "value": "true", - "description": "是否要求所有 action 维度一致", - "type": "string" - }, - { - "key": "MAX_ROWS", - "value": "10000", - "description": "最多读取多少行真实样本", - "type": "string" - }, - { - "key": "MAX_FILES", - "value": "50", - "description": "最多读取多少个清单文件", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/vla_sample_manifest_validate/operator.py", - "runtimeHandler": "builtin_vla_sample_manifest_validate" - }, - { - "id": "operator-builtin-camera-calibration-validate", - "name": "相机标定校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 JSON/YAML 相机标定文件,校验内参、畸变、投影/外参矩阵和图像尺寸。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "camera_info/calibration/intrinsics JSON 或 YAML 文件/目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "相机标定校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向机器人采集、ROS CameraInfo、OpenCV 标定和多相机数据集入库,真实读取 K/D/R/P 或 camera_matrix/distortion/extrinsic 字段,校验矩阵长度、焦距、有限数值和图像尺寸。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "标定 JSON/YAML 文件或包含标定文件的目录", - "type": "string" - }, - { - "key": "MAX_FILES", - "value": "50", - "description": "最多校验多少个标定文件,必须为正整数", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/camera_calibration_validate/operator.py", - "runtimeHandler": "builtin_camera_calibration_validate" - }, - { - "id": "operator-builtin-tf-transform-tree-validate", - "name": "TF 坐标树校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 ROS TransformStamped 风格 JSON/JSONL/YAML 或扁平 CSV 坐标变换,校验父子 frame、平移、四元数归一化、多父节点、重复边和环。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "transforms/tf/tf_static JSON、JSONL、CSV 或 YAML 文件/目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "TF 坐标树校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 ROS/ROS2 机器人数据集、相机外参和多传感器标定入库,真实读取 header.frame_id、child_frame_id、translation、rotation 或 parent_frame/child_frame/qx/qy/qz/qw,校验坐标树根、叶子、重复边、多父节点和环路。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "坐标变换文件或包含坐标变换文件的目录", - "type": "string" - }, - { - "key": "QUATERNION_NORM_TOLERANCE", - "value": "0.01", - "description": "四元数 norm 与 1 的允许误差", - "type": "string" - }, - { - "key": "MAX_TRANSFORMS", - "value": "10000", - "description": "最多校验多少条 transform 记录,必须为正整数", - "type": "string" - }, - { - "key": "MAX_FILES", - "value": "50", - "description": "最多扫描多少个 transform 文件,必须为正整数", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/tf_transform_tree_validate/operator.py", - "runtimeHandler": "builtin_tf_transform_tree_validate" - }, - { - "id": "operator-builtin-ros-joint-state-validate", - "name": "ROS JointState 校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 ROS sensor_msgs/JointState 风格 JSON/JSONL/CSV/Parquet,校验 name、position、velocity、effort 数组长度、有限数值、关节名一致性和可选时间单调性。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "JointState JSON、JSONL、CSV、Parquet 文件或目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "ROS JointState 校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 ROS/ROS2 机器人采集、rosbag 导出、控制日志和训练前关节状态入库,真实读取 sensor_msgs/JointState 风格 name/position/velocity/effort 字段,检查数组长度同 name 一致、数值有限、关节名称序列是否稳定,并可选检查 timestamp 单调。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "JointState 文件或包含 JointState 文件的目录", - "type": "string" - }, - { - "key": "MAX_ROWS", - "value": "100000", - "description": "最多读取多少条 JointState 消息", - "type": "string" - }, - { - "key": "MAX_FILES", - "value": "50", - "description": "最多读取多少个输入文件", - "type": "string" - }, - { - "key": "REQUIRE_CONSISTENT_NAMES", - "value": "true", - "description": "是否要求所有消息的关节名称顺序一致", - "type": "string" - }, - { - "key": "REQUIRE_MONOTONIC_TIME", - "value": "false", - "description": "是否要求 timestamp/header.stamp 单调不降", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/ros_joint_state_validate/operator.py", - "runtimeHandler": "builtin_ros_joint_state_validate" - }, - { - "id": "operator-builtin-graspnet-dataset-validate", - "name": "GraspNet 抓取数据集校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 GraspNet-1Billion 数据根目录、scene 目录或 camera 目录,校验 scenes/scene_XXXX、RGB-D、mask、MAT/XML/rect 标注、object models、grasp_label 和 collision_label。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "GraspNet 根目录、scenes/scene_XXXX 目录或 kinect/realsense camera 目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "GraspNet RGB-D 抓取数据集结构与标注校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 GraspNet-1Billion、AnyGrasp、SuctionNet 和通用物体抓取训练数据,真实读取 models/%03d/nontextured.ply|textured.obj|textured.sdf、grasp_label/%03d_labels.npz、collision_label/scene_XXXX/collision_labels.npz、scenes/scene_XXXX/object_id_list.txt、rs_wrt_kn.npy、kinect|realsense/camK.npy、camera_poses.npy、cam0_wrt_table.npy、rgb/depth/label PNG、meta MAT、annotations XML 和 rect NPY,校验 6DoF grasp label、碰撞标签、相机内外参、RGB-D 帧和实例 mask 是否可用于真实抓取模型训练。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "GraspNet 根目录、scene 目录或 camera 目录", - "type": "string" - }, - { - "key": "CAMERA", - "value": "any", - "description": "校验 camera:any、kinect、realsense 或逗号分隔子集", - "type": "string" - }, - { - "key": "MAX_SCENES", - "value": "20", - "description": "最多校验多少个 scene_XXXX 目录,必须为正整数", - "type": "string" - }, - { - "key": "MAX_FRAMES_PER_CAMERA", - "value": "32", - "description": "每个 camera 最多校验多少帧,必须为正整数", - "type": "string" - }, - { - "key": "REQUIRE_MODELS", - "value": "true", - "description": "是否要求 object_id_list 中每个对象存在 nontextured.ply、textured.obj、textured.sdf", - "type": "boolean" - }, - { - "key": "REQUIRE_GRASP_LABELS", - "value": "true", - "description": "是否要求每个 object id 存在 grasp_label/%03d_labels.npz 且包含 points/offsets/scores/collision", - "type": "boolean" - }, - { - "key": "REQUIRE_COLLISION_LABELS", - "value": "true", - "description": "是否要求每个 scene 存在 collision_label/scene_XXXX/collision_labels.npz", - "type": "boolean" - }, - { - "key": "REQUIRE_RECT_LABELS", - "value": "false", - "description": "是否要求每帧存在 rect/%04d.npy 矩形抓取标签", - "type": "boolean" - }, - { - "key": "REQUIRE_OBJECT_IDS", - "value": "true", - "description": "是否要求 scene 目录存在 object_id_list.txt", - "type": "boolean" - }, - { - "key": "REQUIRE_CAMERA_POSES", - "value": "true", - "description": "是否要求 camera 目录存在 camK.npy、camera_poses.npy、cam0_wrt_table.npy", - "type": "boolean" - }, - { - "key": "REQUIRE_META_MAT", - "value": "true", - "description": "是否要求每帧存在 meta/%04d.mat", - "type": "boolean" - }, - { - "key": "REQUIRE_ANNOTATION_XML", - "value": "true", - "description": "是否要求每帧存在 annotations/%04d.xml 且可解析", - "type": "boolean" - } - ], - "runtimeSource": "backend/operators/builtin/graspnet_dataset_validate/operator.py", - "runtimeHandler": "builtin_graspnet_dataset_validate" - }, - { - "id": "operator-builtin-bop-dataset-validate", - "name": "BOP 6D姿态数据集校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 BOP scenewise 数据根目录、split/scene 目录或 scene_*.json,校验 camera、models_info、scene_camera、scene_gt、scene_gt_info、rgb/depth/mask/mask_visib 与 6D object pose 核心字段。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "BOP 数据根目录、split/scene 目录或单个 scene_*.json 文件", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "BOP 6D object pose 数据集结构与标注校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 BOP benchmark、YCB-V、T-LESS、HOT3D、XYZ-IBD、HOPE 等 6D object pose / RGB-D 机器人感知数据,真实读取 camera.json、models*/models_info.json、train|val|test*//scene_camera*.json、scene_gt*.json、scene_gt_info*.json、rgb|gray、depth、mask、mask_visib,校验 cam_K、depth_scale、obj_id、cam_R_m2c、cam_t_m2c、bbox、visib_fract 等字段并统计 split、scene、image、object 和 annotation 覆盖。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "BOP 数据根目录、split/scene 目录或 scene_*.json 文件", - "type": "string" - }, - { - "key": "MAX_SCENES", - "value": "20", - "description": "最多校验多少个 scene 目录,必须为正整数", - "type": "string" - }, - { - "key": "MAX_IMAGES_PER_SCENE", - "value": "200", - "description": "每个 scene 最多校验多少个 image id,必须为正整数", - "type": "string" - }, - { - "key": "CHECK_IMAGE_FILES", - "value": "true", - "description": "是否要求每个 image id 存在 rgb 或 gray 图像文件", - "type": "boolean" - }, - { - "key": "CHECK_DEPTH_FILES", - "value": "false", - "description": "是否要求每个 image id 存在 depth 图像文件;RGB-only/HOT3D 等数据可关闭", - "type": "boolean" - }, - { - "key": "CHECK_MASK_FILES", - "value": "false", - "description": "是否要求每个 scene_gt annotation 存在 mask 与 mask_visib 文件", - "type": "boolean" - }, - { - "key": "REQUIRE_SCENE_GT", - "value": "true", - "description": "是否要求 scene_gt*.json 存在并包含 object pose 注解", - "type": "boolean" - }, - { - "key": "REQUIRE_SCENE_GT_INFO", - "value": "false", - "description": "是否要求 scene_gt_info*.json 存在并与 scene_gt annotation 数量一致", - "type": "boolean" - }, - { - "key": "REQUIRE_MODELS_INFO", - "value": "true", - "description": "是否要求数据根目录存在 models*/models_info.json", - "type": "boolean" - }, - { - "key": "REQUIRE_MODEL_FILES", - "value": "false", - "description": "是否要求 models_info.json 中每个 obj_id 存在对应 obj_XXXXXX.ply", - "type": "boolean" - }, - { - "key": "REQUIRE_OBJECT_POSE", - "value": "true", - "description": "是否要求每个 annotation 存在 cam_R_m2c 与 cam_t_m2c 且 shape 合法", - "type": "boolean" - } - ], - "runtimeSource": "backend/operators/builtin/bop_dataset_validate/operator.py", - "runtimeHandler": "builtin_bop_dataset_validate" - }, - { - "id": "operator-builtin-dexycb-dataset-validate", - "name": "DexYCB 数据集校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 DexYCB 数据根目录或 subject 子目录,校验 labels_*.npz、color_*.jpg、aligned_depth_to_color_*.png、calibration/models 和手-物体姿态核心数组。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "DexYCB 数据根目录、subject 子目录或单个 labels_*.npz 文件", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "DexYCB RGB-D 与标签结构校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 NVIDIA DexYCB 真实手-物体抓取数据,读取 20200709-subject-01/20200709_141841//labels_000053.npz 这类结构,校验 color_*.jpg、aligned_depth_to_color_*.png、labels_*.npz 中 seg、pose_y、pose_m、joint_3d、joint_2d 的 shape/dtype/finite,统计 subject、camera、object pose 和可见 hand joints,并可要求数据根目录包含 calibration 与 models。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "DexYCB 数据根目录、subject 子目录、camera 目录或单个 labels_*.npz 文件", - "type": "string" - }, - { - "key": "MAX_SAMPLES", - "value": "200", - "description": "最多校验多少个 labels_*.npz 样本,必须为正整数", - "type": "string" - }, - { - "key": "CHECK_COLOR_DEPTH_FILES", - "value": "true", - "description": "是否要求每个 labels_*.npz 对应 color_.jpg 与 aligned_depth_to_color_.png", - "type": "boolean" - }, - { - "key": "REQUIRE_ROOT_ASSETS", - "value": "true", - "description": "是否要求数据根目录包含 calibration/ 与 models/;只校验 subject 分包时可关闭", - "type": "boolean" - }, - { - "key": "REQUIRE_FINITE_NUMERIC", - "value": "true", - "description": "是否检查 label 数组不含 NaN/Inf", - "type": "boolean" - } - ], - "runtimeSource": "backend/operators/builtin/dexycb_dataset_validate/operator.py", - "runtimeHandler": "builtin_dexycb_dataset_validate" - }, - { - "id": "operator-builtin-ho3d-dataset-validate", - "name": "HO3D/H2O3D 数据集校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 HO-3D/H2O-3D 数据根目录、split/sequence 目录或 meta/*.pkl,校验 train/evaluation 下 rgb、depth、segr、meta pickle 和手-物体 3D 姿态核心字段。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "HO-3D/H2O-3D 数据根目录、split/sequence 目录或单个 meta/*.pkl 文件", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "HO-3D/H2O-3D 目录与 meta 注解校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 HO-3D v2/v3 与 H2O-3D 手-物体 3D 姿态数据,真实读取 train//rgb、depth、segr、meta/.pkl 这类结构,校验 camMat、objRot、objTrans、objCorners3DRest、objName、handPose、handTrans、handBeta、handJoints3D、handBoundingBox 和可选 contact map 字段,并统计 split、sequence、object 与手部注解覆盖。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "HO-3D/H2O-3D 数据根目录、split/sequence 目录或单个 meta/*.pkl 文件", - "type": "string" - }, - { - "key": "MAX_SAMPLES", - "value": "200", - "description": "最多校验多少个 meta/*.pkl 样本,必须为正整数", - "type": "string" - }, - { - "key": "CHECK_RGB_DEPTH_FILES", - "value": "true", - "description": "是否要求每个 meta/*.pkl 对应 rgb/.jpg|png 与 depth/.png", - "type": "boolean" - }, - { - "key": "CHECK_SEGMENTATION_FILES", - "value": "false", - "description": "是否要求每个样本存在 segr/.png 分割图", - "type": "boolean" - }, - { - "key": "REQUIRE_OBJECT_POSE", - "value": "true", - "description": "是否要求 objRot 与 objTrans 存在且 shape 合法", - "type": "boolean" - }, - { - "key": "REQUIRE_HAND_ANNOTATION", - "value": "true", - "description": "是否要求存在 MANO handPose/handTrans/handBeta 或 handJoints3D 手部注解", - "type": "boolean" - }, - { - "key": "REQUIRE_FINITE_NUMERIC", - "value": "true", - "description": "是否检查 meta 数组不含 NaN/Inf", - "type": "boolean" - } - ], - "runtimeSource": "backend/operators/builtin/ho3d_dataset_validate/operator.py", - "runtimeHandler": "builtin_ho3d_dataset_validate" - }, - { - "id": "operator-builtin-hand-object-pose-validate", - "name": "手-物体姿态标注校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 DexYCB/HO3D 风格手-物体姿态标注 JSON/JSONL/CSV/Parquet,校验 21 点手部 2D/3D 关键点、物体 6D pose、bbox、图像和 mask 引用。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "手-物体姿态标注文件或数据集目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "手-物体姿态标注校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 DexYCB、HO3D、机器人手抓取和具身操作标注入库,真实读取 image/mask 路径、hand_joints_3d、hand_joints_2d、object_pose_matrix 或 quaternion+translation、bbox 等字段,检查文件存在、关键点 shape、物体位姿和 bbox 合法性。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "标注文件或包含标注文件的目录", - "type": "string" - }, - { - "key": "EXPECTED_HAND_JOINTS", - "value": "21", - "description": "期望手部关键点数量,DexYCB/HO3D 常用 21", - "type": "string" - }, - { - "key": "CHECK_IMAGE_EXISTS", - "value": "true", - "description": "是否要求 image/rgb 文件真实存在", - "type": "string" - }, - { - "key": "CHECK_MASK_EXISTS", - "value": "false", - "description": "是否要求 mask/segmentation 文件真实存在", - "type": "string" - }, - { - "key": "MAX_ROWS", - "value": "100000", - "description": "最多读取多少条姿态标注", - "type": "string" - }, - { - "key": "MAX_FILES", - "value": "50", - "description": "最多读取多少个标注文件", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/hand_object_pose_validate/operator.py", - "runtimeHandler": "builtin_hand_object_pose_validate" - }, - { - "id": "operator-builtin-partnet-mobility-asset-validate", - "name": "PartNet-Mobility 资产校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 PartNet-Mobility/SAPIEN articulated object 资产目录,校验 mobility.urdf、semantics.txt、mobility_v2.json、meta/result/bounding box、textured_objs mesh 与 point_sample。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "PartNet-Mobility 对象目录、dataset/sapien_dataset 根目录或 mobility.urdf 文件", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "PartNet-Mobility/SAPIEN 资产结构、关节树、运动语义和 mesh 引用校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 SAPIEN/PartNet-Mobility、ManiSkill articulated object 和具身仿真数据生成资产入库,真实读取每个对象的 mobility.urdf、semantics.txt、mobility_v2.json、meta.json、result.json、result_after_merging.json、bounding_box.json、textured_objs/*.obj、point_sample 与 parts_render,校验 URDF link/joint 树、hinge/slider/screw/free 运动语义、motion limits/axis、mesh 引用和对象类别统计。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "PartNet-Mobility 对象目录、dataset/sapien_dataset 根目录或 mobility.urdf 文件", - "type": "string" - }, - { - "key": "MAX_OBJECTS", - "value": "50", - "description": "最多校验多少个对象目录,必须为正整数", - "type": "string" - }, - { - "key": "MAX_MESH_REFERENCES", - "value": "200", - "description": "每个对象最多返回多少个 mesh/joint 引用预览", - "type": "string" - }, - { - "key": "REQUIRE_SEMANTICS", - "value": "true", - "description": "是否要求对象存在 semantics.txt 且 link 语义可解析", - "type": "boolean" - }, - { - "key": "REQUIRE_MOBILITY_V2", - "value": "true", - "description": "是否要求对象存在 mobility_v2.json 且包含运动条目", - "type": "boolean" - }, - { - "key": "REQUIRE_META_JSON", - "value": "true", - "description": "是否要求对象存在 meta.json", - "type": "boolean" - }, - { - "key": "REQUIRE_TEXTURED_OBJS", - "value": "true", - "description": "是否要求 textured_objs 下存在真实 .obj mesh", - "type": "boolean" - }, - { - "key": "REQUIRE_RESULT_JSON", - "value": "false", - "description": "是否要求对象存在 result.json", - "type": "boolean" - }, - { - "key": "REQUIRE_POINT_SAMPLE", - "value": "false", - "description": "是否要求 point_sample 下存在点云/标签文件", - "type": "boolean" - }, - { - "key": "REQUIRE_BOUNDING_BOX", - "value": "false", - "description": "是否要求对象存在 bounding_box.json", - "type": "boolean" - } - ], - "runtimeSource": "backend/operators/builtin/partnet_mobility_asset_validate/operator.py", - "runtimeHandler": "builtin_partnet_mobility_asset_validate" - }, - { - "id": "operator-builtin-mjcf-actuator-validate", - "name": "MJCF 执行器校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 MuJoCo/MJCF XML,校验 actuator 下 motor/position/velocity/general 等执行器目标引用、ctrlrange/forcerange/actrange 和 gear 数值。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "MJCF XML 文件或包含 MJCF 文件的目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "MJCF 执行器校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 MuJoCo、robosuite、ManiSkill 和 GitHub 常见 MJCF 仿真资产,真实解析 下执行器,验证 joint/tendon/site/body 引用存在,ctrlrange/forcerange/actrange 为递增有限数值,并识别重复 actuator 名称。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "MJCF XML 文件或包含 MJCF 文件的目录", - "type": "string" - }, - { - "key": "MAX_MODELS", - "value": "20", - "description": "最多扫描多少个 MJCF 文件,必须为正整数", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/mjcf_actuator_validate/operator.py", - "runtimeHandler": "builtin_mjcf_actuator_validate" - }, - { - "id": "operator-builtin-robot-model-asset-validate", - "name": "机器人模型资产校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 URDF/MJCF/XML 机器人模型,校验 link/joint/body/actuator 结构和 mesh/include 文件引用。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "URDF/MJCF/XML 文件或机器人描述包目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "机器人模型资产校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 GitHub 常见 robot_description、URDF/MJCF 仿真资产和内部机器人模型入库,真实解析 XML,统计 link/joint/body/actuator,并解析 mesh filename/file、package:// 和 include 引用,识别缺失资产。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "URDF/MJCF/XML 文件或包含模型文件的目录", - "type": "string" - }, - { - "key": "PACKAGE_ROOTS", - "value": "", - "description": "可选逗号分隔 package 根目录,用于解析 package:// mesh URI", - "type": "string" - }, - { - "key": "MAX_MODELS", - "value": "20", - "description": "最多扫描多少个模型文件,必须为正整数", - "type": "string" - }, - { - "key": "MAX_REFERENCES", - "value": "100", - "description": "每个模型最多返回多少个 mesh/include 引用预览", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/robot_model_asset_validate/operator.py", - "runtimeHandler": "builtin_robot_model_asset_validate" - }, - { - "id": "operator-builtin-joint-limit-validate", - "name": "URDF 关节限位校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "解析真实 URDF joint limit,并读取 JSON/JSONL/CSV/Parquet 轨迹中的 qpos/joint_positions,校验关节位置是否越界。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "包含 URDF 和轨迹文件的目录,或通过 MODEL_PATH/TRAJECTORY_PATH 指定真实文件", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "URDF 关节限位校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向真实机器人描述包和采集轨迹,解析 URDF revolute/prismatic joint 的 lower/upper limit,读取 joint_positions、observation.joint_positions、state 或 qpos 字段,按 JOINT_NAMES 或 URDF 顺序映射,发现维度不匹配和超限样本。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "包含 URDF 与轨迹文件的目录", - "type": "string" - }, - { - "key": "MODEL_PATH", - "value": "", - "description": "可选,显式 URDF 文件路径", - "type": "string" - }, - { - "key": "TRAJECTORY_PATH", - "value": "", - "description": "可选,显式 JSON/JSONL/CSV/Parquet 轨迹文件或目录", - "type": "string" - }, - { - "key": "POSITION_FIELDS", - "value": "joint_positions,observation.joint_positions,observation.state,state,states,qpos", - "description": "逗号分隔的候选关节位置字段,支持点路径", - "type": "string" - }, - { - "key": "JOINT_NAME_FIELDS", - "value": "joint_names,observation.joint_names,names", - "description": "逗号分隔的候选关节名字段,支持点路径", - "type": "string" - }, - { - "key": "JOINT_NAMES", - "value": "", - "description": "可选,逗号分隔的轨迹列顺序;为空时使用 URDF 中有上下限的关节顺序", - "type": "string" - }, - { - "key": "TOLERANCE", - "value": "0", - "description": "允许的限位浮点容差", - "type": "string" - }, - { - "key": "MAX_ROWS", - "value": "100000", - "description": "最多读取多少行真实轨迹", - "type": "string" - }, - { - "key": "MAX_FILES", - "value": "50", - "description": "最多读取多少个轨迹文件", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/joint_limit_validate/operator.py", - "runtimeHandler": "builtin_joint_limit_validate" - }, - { - "id": "operator-builtin-pointcloud-summary", - "name": "点云数据概览", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 PCD/PLY 点云文件,输出点数、字段、编码和 ASCII 点云空间边界。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "PCD/PLY 文件或目录输入", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "summary", - "type": "json", - "description": "点云文件概览", - "transmission": "bus" - } - ], - "skillDefinition": "面向激光雷达、深度相机重建和具身空间感知数据,真实读取 PCD/PLY header、fields/properties、point count 和 ASCII x/y/z 采样边界,不把点云当普通文件名扫描。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": ".pcd/.ply 文件或包含点云文件的目录", - "type": "string" - }, - { - "key": "MAX_FILES", - "value": "50", - "description": "最多扫描多少个点云文件,必须为正整数", - "type": "string" - }, - { - "key": "MAX_SAMPLE_POINTS", - "value": "200000", - "description": "每个 ASCII 点云最多采样多少点计算边界", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/pointcloud_summary/operator.py", - "runtimeHandler": "builtin_pointcloud_summary" - }, - { - "id": "operator-builtin-sensor-timestamp-sync-validate", - "name": "多传感器时间同步校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 CSV/JSONL 时间戳清单,校验 camera/lidar/action 等多传感器流的同步完整性和最大漂移。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "包含 stream/timestamp/frame_id 的 CSV 或 JSONL 时间戳文件", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "多传感器时间同步校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向机器人多相机、LiDAR、action/state 采集链路,真实读取 timestamp/stamp、timestamp_ns/time_ns、timestamp_ms/time_ms、ROS sec+nsec 字段,按 frame_id/sequence 或最近邻分组,计算同步组缺失流和最大时间漂移。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "CSV/JSONL 时间戳文件或包含时间戳文件的目录", - "type": "string" - }, - { - "key": "REQUIRED_STREAMS", - "value": "camera,lidar,action", - "description": "逗号分隔的必需传感器流名称", - "type": "string" - }, - { - "key": "MAX_DRIFT_MS", - "value": "50", - "description": "单个同步组允许的最大时间漂移毫秒数", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/sensor_timestamp_sync_validate/operator.py", - "runtimeHandler": "builtin_sensor_timestamp_sync_validate" - }, - { - "id": "operator-builtin-lerobot-dataset-validate", - "name": "LeRobot 数据集校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 LeRobot v2/v3 数据集目录和 Parquet 数据文件,校验 meta、episode、schema、frame index 和视频引用。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "LeRobot 数据集根目录或上游目录 artifact", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "LeRobot 数据集校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 Hugging Face LeRobot 数据集真实落地验收,读取 meta/info.json、meta/episodes/*.parquet 或 episodes.jsonl、tasks.jsonl、stats.json 和 data/*.parquet,校验 schema 与 frame/episode 计数。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "LeRobot 数据集根目录或其中任意文件路径", - "type": "string" - }, - { - "key": "REQUIRED_COLUMNS", - "value": "episode_index,frame_index,timestamp,action", - "description": "Parquet 必需列,逗号分隔", - "type": "string" - }, - { - "key": "MAX_PARQUET_FILES", - "value": "24", - "description": "最多读取多少个 Parquet 文件的 schema 和索引", - "type": "string" - }, - { - "key": "MAX_INDEX_ROWS", - "value": "200000", - "description": "最多扫描多少行 episode_index/frame_index 以检查顺序", - "type": "string" - }, - { - "key": "CHECK_INDEX_ORDER", - "value": "true", - "description": "是否检查每个 episode 内 frame_index 单调不降", - "type": "string" - }, - { - "key": "CHECK_VIDEOS", - "value": "true", - "description": "当 info.json 声明 total_videos 时是否要求目录里存在真实视频文件", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/lerobot_dataset_validate/operator.py", - "runtimeHandler": "builtin_lerobot_dataset_validate" - }, - { - "id": "operator-builtin-lerobot-video-alignment-validate", - "name": "LeRobot 视频对齐校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 LeRobot v3 data Parquet、meta/episodes 和 MP4 视频文件,校验声明视频数、视频帧数、数据行数、episode length 和显式视频引用是否一致。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "LeRobot 数据集根目录或上游目录 artifact", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "LeRobot 视频对齐校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 Hugging Face LeRobot v3 数据集、远程机器人视频轨迹和训练前数据落地验收,真实读取 meta/info.json 的 video feature、meta/episodes、data Parquet 和 videos 目录,使用 OpenCV 读取 MP4 帧数/FPS/尺寸,检查 Parquet 行数、episode 长度、视频文件数和视频帧数是否对齐。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "LeRobot 数据集根目录或其中任意文件路径", - "type": "string" - }, - { - "key": "MAX_PARQUET_FILES", - "value": "50", - "description": "最多读取多少个 meta/data Parquet 文件", - "type": "string" - }, - { - "key": "MAX_ROWS", - "value": "200000", - "description": "最多扫描多少行真实 Parquet 数据以统计 episode 和视频引用", - "type": "string" - }, - { - "key": "MAX_VIDEOS", - "value": "200", - "description": "最多读取多少个视频文件的帧数与基础属性", - "type": "string" - }, - { - "key": "FRAME_TOLERANCE", - "value": "0", - "description": "视频帧数与数据行数允许误差,必须为非负整数", - "type": "string" - }, - { - "key": "REQUIRE_VIDEOS", - "value": "true", - "description": "是否要求存在真实视频文件", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/lerobot_video_alignment_validate/operator.py", - "runtimeHandler": "builtin_lerobot_video_alignment_validate" - }, - { - "id": "operator-builtin-lerobot-metadata-consistency-validate", - "name": "LeRobot 元数据一致性校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 LeRobot v3 meta/info、stats、tasks、episodes 与 data parquet,校验声明计数、feature、stats 和 episode length 一致性。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "LeRobot 数据集根目录或上游目录 artifact", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "LeRobot 元数据一致性校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 Hugging Face LeRobot v3 数据集落地验收,真实读取 meta/info.json 的 features/total_*、meta/stats.json、meta/tasks.jsonl、meta/episodes Parquet/JSONL 与 data Parquet,检查 action/observation.state 等训练必需 feature 是否声明、是否有 stats、是否真的出现在数据列中,以及 episode length 和 task 索引是否一致。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "LeRobot 数据集根目录或其中任意文件路径", - "type": "string" - }, - { - "key": "REQUIRED_FEATURES", - "value": "action,observation.state", - "description": "必需 feature/数据列,逗号分隔", - "type": "string" - }, - { - "key": "MAX_PARQUET_FILES", - "value": "50", - "description": "最多读取多少个 meta/data Parquet 文件", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/lerobot_metadata_consistency_validate/operator.py", - "runtimeHandler": "builtin_lerobot_metadata_consistency_validate" - }, - { - "id": "operator-builtin-lerobot-v3-episode-index-validate", - "name": "LeRobot v3 Episode索引校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 LeRobot v3 meta/episodes 与 data Parquet,校验 episode metadata 的 length、offset、chunk/data path template 与实际 data shard 行范围是否一致。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "LeRobot v3 数据集根目录或上游目录 artifact", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "LeRobot v3 episode 索引校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 Hugging Face LeRobot v3 大规模具身数据集,真实读取 meta/info.json、meta/episodes/*.parquet 和 data/**/*.parquet,验证 data_path template、chunk_index、episode length、data row window 与实际 Parquet shard 映射是否一致,补齐训练前对 episode 索引错位和 shard 漏行的验收。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "LeRobot v3 数据集根目录或其中任意文件路径", - "type": "string" - }, - { - "key": "MAX_PARQUET_FILES", - "value": "100", - "description": "最多扫描多少个 episode/data Parquet 文件", - "type": "string" - }, - { - "key": "CHUNK_SIZE", - "value": "1000", - "description": "未显式提供 chunk_index 时用于 episode_index 推导 chunk 的大小", - "type": "string" - }, - { - "key": "DATA_PATH_TEMPLATE", - "value": "", - "description": "可选 data path 模板,例如 data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", - "type": "string" - }, - { - "key": "STRICT_PATH_MATCH", - "value": "true", - "description": "候选 data path 存在时是否要求该文件实际包含对应 episode 行", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/lerobot_v3_episode_index_validate/operator.py", - "runtimeHandler": "builtin_lerobot_v3_episode_index_validate" - }, - { - "id": "operator-builtin-lerobot-summary", - "name": "LeRobot 数据集概览", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 LeRobot 数据集目录,输出 meta/tasks/parquet/videos 概览。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "目录输入", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "summary", - "type": "json", - "description": "概览结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 LeRobot 数据集目录的真实概览算子,适合机器人示教数据检查。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "LeRobot 数据集根目录", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/lerobot_summary/operator.py", - "runtimeHandler": "builtin_lerobot_summary" - }, - { - "id": "operator-builtin-rosbag-summary", - "name": "ROSBag 数据概览", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 rosbag1 / rosbag2 / mcap / db3 数据并输出概览。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "bag 输入", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "summary", - "type": "json", - "description": "概览结果", - "transmission": "bus" - } - ], - "skillDefinition": "支持 rosbag db3 topic/message 概览,以及 bag/mcap 文件摘要。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "bag/db3/mcap 文件或 rosbag2 目录", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/rosbag_summary/operator.py", - "runtimeHandler": "builtin_rosbag_summary" - }, - { - "id": "operator-builtin-mcap-topic-summary", - "name": "MCAP 主题概览", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 MCAP 文件,解析 schema、channel、topic 和 message payload 统计。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "MCAP 文件或目录输入", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "summary", - "type": "json", - "description": "MCAP topic/schema/message 概览", - "transmission": "bus" - } - ], - "skillDefinition": "面向 ROS2/机器人传感器日志的真实 MCAP 数据,使用 mcap runtime 解析 topic、schema、channel、message payload bytes 和时间范围,而不是只读取文件头。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": ".mcap 文件或包含 .mcap 文件的目录", - "type": "string" - }, - { - "key": "MAX_MESSAGES", - "value": "200000", - "description": "最多扫描多少条 message,必须为正整数", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/mcap_topic_summary/operator.py", - "runtimeHandler": "builtin_mcap_topic_summary" - }, - { - "id": "operator-builtin-tfrecord-structural-validate", - "name": "TFRecord 结构校验", - "type": "builtin", - "category": "ROBOTICS", - "description": "读取真实 TFRecord/RLDS shard,校验 record framing、长度、CRC32C,并输出样本预览。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "TFRecord 文件或目录输入", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "TFRecord 结构校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 RLDS/Open X-Embodiment 等 TensorFlow 数据链路,真实读取 TFRecord 二进制 record framing,校验 length/header CRC 和 payload CRC,识别截断或损坏 shard。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": ".tfrecord/.tfrecords/.tfrec 文件或包含这些文件的目录", - "type": "string" - }, - { - "key": "MAX_RECORDS", - "value": "200000", - "description": "最多扫描多少条 record,必须为正整数", - "type": "string" - }, - { - "key": "PREVIEW_RECORDS", - "value": "5", - "description": "最多返回多少条 record 预览,必须为正整数", - "type": "string" - }, - { - "key": "VALIDATE_CRC", - "value": "true", - "description": "是否校验 TFRecord masked CRC32C", - "type": "string" - } - ], - "runtimeSource": "backend/operators/builtin/tfrecord_structural_validate/operator.py", - "runtimeHandler": "builtin_tfrecord_structural_validate" - }, - { - "id": "operator-builtin-dataset-split-leakage-validate", - "name": "数据集切分泄漏校验", - "type": "builtin", - "category": "DATASET", - "description": "读取真实 CSV/JSON/JSONL/Parquet 切分清单,校验 train/val/test 等 split 之间是否存在 episode 泄漏、缺失 split 或重复样本。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "可选上游数据;通常使用 SOURCE_PATH 指向真实切分清单或数据集目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "切分泄漏校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向具身训练集发布和验收,确保 episode 不会同时出现在 train/val/test 等不同 split 中,并统计每个任务在各 split 的覆盖情况。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "真实 split manifest 文件或包含 split 清单的数据集目录", - "type": "string" - }, - { - "key": "REQUIRED_SPLITS", - "value": "train,val,test", - "description": "必须存在的 split 名称,逗号分隔;可按业务改为 train,val 或其它命名", - "type": "string" - }, - { - "key": "MIN_ROWS_PER_SPLIT", - "value": "1", - "description": "每个必需 split 的最小样本数", - "type": "number" - }, - { - "key": "EPISODE_ID_FIELDS", - "value": "episodeId,episode_id,episode_index,episodeIndex,trajectory_id,trajectoryId,demoId,id,name", - "description": "候选 episode ID 字段,逗号分隔", - "type": "string" - }, - { - "key": "SPLIT_FIELDS", - "value": "split,subset,set,fold", - "description": "候选 split 字段,逗号分隔", - "type": "string" - }, - { - "key": "TASK_FIELDS", - "value": "task,taskName,task_index,taskIndex,language_instruction,instruction", - "description": "候选任务字段,用于统计任务在各 split 的分布", - "type": "string" - }, - { - "key": "MAX_FILES", - "value": "50", - "description": "最多扫描的清单文件数", - "type": "number" - }, - { - "key": "MAX_ROWS", - "value": "200000", - "description": "最多扫描的清单行数", - "type": "number" - } - ], - "runtimeSource": "backend/operators/builtin/dataset_split_leakage_validate/operator.py", - "runtimeHandler": "builtin_dataset_split_leakage_validate" - }, - { - "id": "operator-builtin-media-reference-validate", - "name": "媒体引用可读性校验", - "type": "builtin", - "category": "DATASET", - "description": "读取真实 CSV/JSON/JSONL/Parquet manifest 中的图片、视频、深度图、mask、点云引用,检查文件存在并可被对应解码器读取。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "可选上游数据;通常使用 SOURCE_PATH 指向真实 manifest 或数据集目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "媒体引用校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "面向 HuggingFace/GitHub 常见 manifest + media 文件组织,防止训练前才发现图片、视频、深度帧或点云路径缺失、不可读。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "真实 manifest 文件或包含 manifest 与媒体文件的数据集目录", - "type": "string" - }, - { - "key": "MEDIA_FIELDS", - "value": "image,image_path,imagePath,video,video_path,videoPath,rgb,rgb_path,depth,depth_path,mask,mask_path,pointcloud,pointcloud_path,observation.image,observation.rgb,observation.depth,observation.video,observation.images.rgb,observation.images.wrist", - "description": "显式媒体引用字段,支持点路径,逗号分隔", - "type": "string" - }, - { - "key": "INFER_FIELDS", - "value": "true", - "description": "是否额外从字段名包含 image/video/rgb/depth/mask/path 等关键词的列中推断媒体引用", - "type": "boolean" - }, - { - "key": "DEEP_CHECK", - "value": "true", - "description": "是否读取并解码图片、深度数组、视频元数据或点云文件头", - "type": "boolean" - }, - { - "key": "MAX_FILES", - "value": "50", - "description": "最多扫描的 manifest 文件数", - "type": "number" - }, - { - "key": "MAX_ROWS", - "value": "200000", - "description": "最多扫描的 manifest 行数", - "type": "number" - }, - { - "key": "MAX_REFERENCES", - "value": "5000", - "description": "最多检查的媒体引用数量", - "type": "number" - } - ], - "runtimeSource": "backend/operators/builtin/media_reference_validate/operator.py", - "runtimeHandler": "builtin_media_reference_validate" - }, - { - "id": "operator-builtin-dataset-schema-contract-validate", - "name": "数据集Schema合约校验", - "type": "builtin", - "category": "DATASET", - "description": "读取真实 CSV/JSON/JSONL/Parquet/HDF5/NPZ 文件,按必需列、HDF5 dataset、NPZ array 合约校验训练输入 schema。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "可选上游数据;通常使用 SOURCE_PATH 指向真实数据文件或目录", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "validation", - "type": "json", - "description": "Schema 合约校验结果", - "transmission": "bus" - } - ], - "skillDefinition": "把具身训练前的 schema 假设显式化,统一检查表格列、HDF5 dataset 与 NPZ array 是否存在且非空。", - "hyperparams": [ - { - "key": "SOURCE_PATH", - "value": "", - "description": "真实 CSV/JSON/JSONL/Parquet/HDF5/NPZ 文件或目录", - "type": "string" - }, - { - "key": "REQUIRED_COLUMNS", - "value": "", - "description": "表格类文件必须存在的列或字段,逗号分隔;可用 CONTRACT_JSON.requiredColumns 替代", - "type": "string" - }, - { - "key": "REQUIRED_HDF5_DATASETS", - "value": "", - "description": "HDF5 文件必须存在的 dataset 路径,逗号分隔;可用 CONTRACT_JSON.requiredHdf5Datasets 替代", - "type": "string" - }, - { - "key": "REQUIRED_NPZ_ARRAYS", - "value": "", - "description": "NPZ 文件必须存在的 array 名称,逗号分隔;可用 CONTRACT_JSON.requiredNpzArrays 替代", - "type": "string" - }, - { - "key": "CONTRACT_JSON", - "value": "", - "description": "可选 JSON 合约对象,支持 requiredColumns、requiredHdf5Datasets、requiredNpzArrays", + "transmission": "bus", "type": "json" }, { - "key": "MAX_FILES", - "value": "50", - "description": "最多扫描的数据文件数", - "type": "number" - }, - { - "key": "MAX_ROWS", - "value": "10000", - "description": "表格类文件最多采样行数", - "type": "number" - } - ], - "runtimeSource": "backend/operators/builtin/dataset_schema_contract_validate/operator.py", - "runtimeHandler": "builtin_dataset_schema_contract_validate" - }, - { - "id": "operator-builtin-vla-model-train", - "name": "轻量线性策略训练", - "type": "builtin", - "category": "AI", - "description": "基于真实 VLA 样本清单训练轻量线性策略模型,作为 CPU 可运行的具身策略基线。", - "inputs": [ - { - "id": "in-1", - "name": "samples", - "type": "json", - "description": "包含 instruction 与 action 的 VLA 样本清单或上游样本 artifact", - "transmission": "bus" + "description": "右输入", + "id": "in-2", + "name": "right", + "transmission": "bus", + "type": "json" } ], + "name": "数据合并", "outputs": [ { + "description": "合并结果", "id": "out-1", - "name": "model", - "type": "json", - "description": "训练后的 VLA 策略模型与训练指标", - "transmission": "bus" + "name": "merged", + "transmission": "bus", + "type": "json" } ], - "skillDefinition": "读取真实 VLA 样本并拟合可复现的线性策略基线;缺少真实 action 时失败。该算子不冒充 ACT/OpenVLA 等深度模型训练。", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_merge", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-merge", + "operatorSlug": "merge", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-artifacts" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_merge", + "runtimeSource": "backend/operators/builtin/merge/operator.py", + "skillDefinition": "对多个 JSON 输入执行真实合并。", + "type": "builtin" + }, + { + "category": "IO", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "收集上游真实输出并生成上传结果清单。", + "id": "operator-builtin-upload", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-upload:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-upload@sha256:bc56ba702686c5ed78cc2ab77ede4fa2476acba81df1aa1bd3ea5d6c7ef88474", + "inputs": [ + { + "description": "待上传结果", + "id": "in-1", + "name": "result", + "transmission": "bus", + "type": "json" + } + ], + "name": "结果上传", + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_upload", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-upload", + "operatorSlug": "upload", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "filesystem", + "optional-baidu-bos" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_upload", + "runtimeSource": "backend/operators/builtin/upload/operator.py", + "skillDefinition": "复制或汇总上游产物,生成上传结果清单。", + "type": "builtin" + }, + { + "category": "HUMAN", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "将真实输入送入人工审核并等待审核动作恢复流程。", + "id": "operator-builtin-human-review", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-human-review:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-human-review@sha256:8e789d73f7933102ce12364c2a382affaa16639491393abfdd3092bc3090215d", + "inputs": [ + { + "description": "候选数据", + "id": "in-1", + "name": "candidate", + "transmission": "bus", + "type": "json" + } + ], + "name": "人工审核", + "outputs": [ + { + "description": "审核通过", + "id": "approved", + "name": "approved", + "transmission": "bus", + "type": "json" + }, + { + "description": "审核拒绝", + "id": "rejected", + "name": "rejected", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_human_review", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-human-review", + "operatorSlug": "human_review", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-artifacts" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_human_review", + "runtimeSource": "backend/operators/builtin/human_review/operator.py", + "skillDefinition": "基于真实输入生成审核任务并等待人工处理。", + "type": "builtin" + }, + { + "category": "IO", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "对真实目录做文件统计、扩展名分布和样例文件扫描。", "hyperparams": [ { + "description": "待扫描目录", "key": "SOURCE_PATH", - "value": "", - "description": "可选样本清单路径,支持 JSON/JSONL/CSV 或目录", - "type": "string" - }, - { - "key": "MODEL_NAME", - "value": "vla-linear-policy", - "description": "输出模型名称", - "type": "string" - }, - { - "key": "TEXT_FIELDS", - "value": "", - "description": "语言指令字段候选,逗号分隔", - "type": "string" - }, - { - "key": "ACTION_FIELDS", - "value": "", - "description": "动作字段候选,逗号分隔", - "type": "string" - }, - { - "key": "MIN_SAMPLES", - "value": "2", - "description": "最少训练样本数", - "type": "number" + "type": "string", + "value": "" + } + ], + "id": "operator-builtin-directory-scan", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-directory-scan:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-directory-scan@sha256:f591c68e5a9f63ed1203347dedfddfe7f9f8786266d69af1fbdab74bc4eece9f", + "inputs": [ + { + "description": "目录输入", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "目录扫描", + "outputs": [ + { + "description": "目录概览", + "id": "out-1", + "name": "summary", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_directory_scan", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-directory-scan", + "operatorSlug": "directory_scan", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "filesystem" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_directory_scan", + "runtimeSource": "backend/operators/builtin/directory_scan/operator.py", + "skillDefinition": "读取 SOURCE_PATH 或上游 source manifest,输出真实目录概览。", + "type": "builtin" + }, + { + "category": "DATA", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "将真实 CSV/TSV 文件转换成结构化 JSON 行列表。", + "hyperparams": [ + { + "description": "csv/tsv 文件路径", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + } + ], + "id": "operator-builtin-csv-to-json", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-csv-to-json:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-csv-to-json@sha256:9137bbc331c43d783ac20b3316c606612d96c1e725c95bfdac02f6de6d163c7f", + "inputs": [ + { + "description": "CSV/TSV 输入", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "CSV 转 JSON", + "outputs": [ + { + "description": "JSON 行列表", + "id": "out-1", + "name": "table", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_csv_to_json", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-csv-to-json", + "operatorSlug": "csv_to_json", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "csv-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_csv_to_json", + "runtimeSource": "backend/operators/builtin/csv_to_json/operator.py", + "skillDefinition": "读取 SOURCE_PATH 或上游 source manifest 指向的 csv/tsv 文件,输出完整 JSON 行数据。", + "type": "builtin" + }, + { + "category": "DATA", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "将 JSON 行数据导出为真实 CSV 文件。", + "id": "operator-builtin-json-to-csv", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-json-to-csv:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-json-to-csv@sha256:17d52fd86f00c899e777859a4eef6fb4e16496cbdadbf085ec51c5e79b8bb724", + "inputs": [ + { + "description": "JSON 行列表", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "JSON 转 CSV", + "outputs": [ + { + "description": "CSV 文件", + "id": "out-1", + "name": "csv", + "transmission": "bus", + "type": "file" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_json_to_csv", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-json-to-csv", + "operatorSlug": "json_to_csv", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-artifacts" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_json_to_csv", + "runtimeSource": "backend/operators/builtin/json_to_csv/operator.py", + "skillDefinition": "把上游 JSON items/entries 或单条对象导出为真实 CSV 文件,并附带摘要。", + "type": "builtin" + }, + { + "category": "DATA", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "按字段路径从 JSON 输入中提取指定值。", + "hyperparams": [ + { + "description": "点分字段路径,例如 metrics.score", + "key": "FIELD_PATH", + "type": "string", + "value": "" + } + ], + "id": "operator-builtin-json-field-extract", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-json-field-extract:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-json-field-extract@sha256:f7b6df195863eef04846a3372f02cb5cac149052496976c65891d7cb3f844945", + "inputs": [ + { + "description": "JSON 输入", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "JSON 字段提取", + "outputs": [ + { + "description": "提取结果", + "id": "out-1", + "name": "extracted", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_json_extract", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-json-field-extract", + "operatorSlug": "json_field_extract", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-artifacts" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_json_extract", + "runtimeSource": "backend/operators/builtin/json_field_extract/operator.py", + "skillDefinition": "根据 FIELD_PATH 从 JSON 输入中提取字段。", + "type": "builtin" + }, + { + "category": "DATA", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "按字段条件过滤 JSON 列表。", + "hyperparams": [ + { + "description": "待判断字段路径", + "key": "FIELD_PATH", + "type": "string", + "value": "" + }, + { + "description": "eq/ne/contains/gt/gte/lt/lte/exists/missing/in;显式覆盖时必须是受支持值", + "key": "OPERATOR", + "type": "string", + "value": "eq" + }, + { + "description": "字符串比较值", + "key": "EXPECTED", + "type": "string", + "value": "" + }, + { + "description": "JSON 格式比较值,优先于 EXPECTED", + "key": "EXPECTED_JSON", + "type": "string", + "value": "" + } + ], + "id": "operator-builtin-json-filter", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-json-filter:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-json-filter@sha256:824f00ea594090923c52d78d0b649abef1b47d37e0d8f1211734c39c66943bbb", + "inputs": [ + { + "description": "JSON 输入", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "JSON 过滤", + "outputs": [ + { + "description": "过滤结果", + "id": "out-1", + "name": "filtered", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_json_filter", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-json-filter", + "operatorSlug": "json_filter", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-artifacts" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_json_filter", + "runtimeSource": "backend/operators/builtin/json_filter/operator.py", + "skillDefinition": "按 FIELD_PATH + OPERATOR + EXPECTED 条件真实过滤 JSON items。", + "type": "builtin" + }, + { + "category": "DATA", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "按映射表批量重命名 JSON 字段。", + "hyperparams": [ + { + "description": "JSON 字段映射,例如 {\"old\":\"new\"};源/目标字段必须是非空字符串且不得导致输出字段覆盖", + "key": "FIELD_MAP_JSON", + "type": "string", + "value": "{}" + } + ], + "id": "operator-builtin-json-map-rename", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-json-map-rename:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-json-map-rename@sha256:53a425b6fabacbf148a17aaf411ce6762c9c5dcd9f4bdf2802edd9f12fc18c72", + "inputs": [ + { + "description": "JSON 输入", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "JSON 映射重命名", + "outputs": [ + { + "description": "映射结果", + "id": "out-1", + "name": "mapped", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_json_map", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-json-map-rename", + "operatorSlug": "json_map_rename", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-artifacts" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_json_map", + "runtimeSource": "backend/operators/builtin/json_map_rename/operator.py", + "skillDefinition": "根据 FIELD_MAP_JSON 真实执行字段名重映射;源字段和目标字段必须是非空字符串,且不能把多个输入字段写到同一输出字段。", + "type": "builtin" + }, + { + "category": "UTIL", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "使用上游 JSON 上下文渲染真实文本模板文件。", + "hyperparams": [ + { + "description": "模板文本", + "key": "TEMPLATE_TEXT", + "type": "string", + "value": "" + }, + { + "description": "输出文件名", + "key": "OUTPUT_NAME", + "type": "string", + "value": "rendered.txt" + } + ], + "id": "operator-builtin-template-render", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-template-render:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-template-render@sha256:b96fcf95d3500eb6849cdb61df7a5ae6f4d3266fc5e69d6b9cce6ce2a05c8fc4", + "inputs": [ + { + "description": "JSON 输入", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "文本模板渲染", + "outputs": [ + { + "description": "渲染后的文本文件", + "id": "out-1", + "name": "rendered", + "transmission": "bus", + "type": "file" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_text_template", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-template-render", + "operatorSlug": "template_render", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-artifacts" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_text_template", + "runtimeSource": "backend/operators/builtin/template_render/operator.py", + "skillDefinition": "支持 {{field}} 和 Python format 风格占位符,输出真实文本文件。", + "type": "builtin" + }, + { + "category": "LOGIC", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "按字段条件将输入拆分到 matched / unmatched 两个真实输出端口。", + "hyperparams": [ + { + "description": "待判断字段路径", + "key": "FIELD_PATH", + "type": "string", + "value": "" + }, + { + "description": "eq/ne/contains/gt/gte/lt/lte/exists/missing/in;显式覆盖时必须是受支持值", + "key": "OPERATOR", + "type": "string", + "value": "eq" + }, + { + "description": "字符串比较值", + "key": "EXPECTED", + "type": "string", + "value": "" + }, + { + "description": "JSON 格式比较值,优先于 EXPECTED", + "key": "EXPECTED_JSON", + "type": "string", + "value": "" + } + ], + "id": "operator-builtin-conditional-route", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-conditional-route:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-conditional-route@sha256:0de386aacb81c0658bac316e50689d3fa56bc7403865ce3237262e1dda0ff5a4", + "inputs": [ + { + "description": "JSON 输入", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "条件分流", + "outputs": [ + { + "description": "命中规则的数据", + "id": "matched", + "name": "matched", + "transmission": "bus", + "type": "json" + }, + { + "description": "未命中规则的数据", + "id": "unmatched", + "name": "unmatched", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_condition_router", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-conditional-route", + "operatorSlug": "conditional_route", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-artifacts" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_condition_router", + "runtimeSource": "backend/operators/builtin/conditional_route/operator.py", + "skillDefinition": "按 FIELD_PATH + OPERATOR + EXPECTED 条件把真实数据分流到两个输出端口。", + "type": "builtin" + }, + { + "category": "DATASET", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "扫描 X-platform 采集目录,解析 meta.json、video_meta.json 和视频文件索引。", + "hyperparams": [ + { + "description": "X-platform 根目录或单个采集目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + } + ], + "id": "operator-builtin-xplatform-dataset-index", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-xplatform-dataset-index:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-xplatform-dataset-index@sha256:3dacff68fa10e013999fc940605d2f318d6422e941be392bea05ada8ca3d7908", + "inputs": [ + { + "description": "目录输入", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "X 平台数据集索引", + "outputs": [ + { + "description": "数据集索引", + "id": "out-1", + "name": "dataset", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_xplatform_index", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-xplatform-dataset-index", + "operatorSlug": "xplatform_dataset_index", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "xplatform-dataset-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_xplatform_index", + "runtimeSource": "backend/operators/builtin/xplatform_dataset_index/operator.py", + "skillDefinition": "面向 BJ_***/DJI_***/meta.json/video_meta.json 结构的真实数据集索引算子,输出采集目录和视频清单。", + "type": "builtin" + }, + { + "category": "DATASET", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "校验采集目录是否包含 meta.json、video_meta.json,以及可选 intrinsics.json。", + "hyperparams": [ + { + "description": "是否要求 intrinsics.json 必须存在", + "key": "REQUIRE_INTRINSICS", + "type": "string", + "value": "false" + } + ], + "id": "operator-builtin-xplatform-dataset-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-xplatform-dataset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-xplatform-dataset-validate@sha256:10a3a8c6a775768205986dd9d81e938c5669be3ada995d03910cc28a3c58aaf3", + "inputs": [ + { + "description": "数据集索引", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "采集目录校验", + "outputs": [ + { + "description": "校验结果", + "id": "out-1", + "name": "validated", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_collection_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-xplatform-dataset-validate", + "operatorSlug": "xplatform_dataset_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "xplatform-dataset-index" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_collection_validate", + "runtimeSource": "backend/operators/builtin/xplatform_dataset_validate/operator.py", + "skillDefinition": "对 X-platform 数据集索引结果执行真实结构校验,输出每个采集目录的有效性报告。", + "type": "builtin" + }, + { + "category": "VIDEO", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "基于真实视频帧计算清晰度、抖动度和无手率估计。", + "hyperparams": [ + { + "description": "清晰度最低阈值", + "key": "MIN_CLARITY", + "type": "string", + "value": "0.8" + }, + { + "description": "抖动度最高阈值", + "key": "MAX_SHAKE", + "type": "string", + "value": "0.1" + }, + { + "description": "无手率最高阈值", + "key": "MAX_HAND_OCCLUSION", + "type": "string", + "value": "0.1" + }, + { + "description": "清晰度归一化尺度,显式覆盖时必须为大于 0 的有限数字", + "key": "CLARITY_SCALE", + "type": "string", + "value": "35" + } + ], + "id": "operator-builtin-video-quality", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-quality:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-quality@sha256:755209a0ec2c869a9784e70fa4cad0803dda72f031791f1e43363bc38b3c4ece", + "inputs": [ + { + "description": "视频清单", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "视频质检", + "outputs": [ + { + "description": "质检结果", + "id": "out-1", + "name": "quality", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_video_quality", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-video-quality", + "operatorSlug": "video_quality", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "ffprobe", + "video-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_video_quality", + "runtimeSource": "backend/operators/builtin/video_quality/operator.py", + "skillDefinition": "使用 OpenCV 对真实 mp4 做质检分析,输出 clarity / shake / handOcclusion / passed。", + "type": "builtin" + }, + { + "category": "VIDEO", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "按日期、地点、场景和质检阈值过滤有效视频。", + "hyperparams": [ + { + "description": "开始日期,支持 YYYY-MM-DD / YYYYMMDD", + "key": "DATE_START", + "type": "string", + "value": "" + }, + { + "description": "结束日期,支持 YYYY-MM-DD / YYYYMMDD", + "key": "DATE_END", + "type": "string", + "value": "" + }, + { + "description": "采集地点", + "key": "PLACE", + "type": "string", + "value": "" + }, + { + "description": "采集场景类型", + "key": "TYPE", + "type": "string", + "value": "" + }, + { + "description": "清晰度最低阈值", + "key": "MIN_CLARITY", + "type": "string", + "value": "0.8" + }, + { + "description": "抖动度最高阈值", + "key": "MAX_SHAKE", + "type": "string", + "value": "0.1" + }, + { + "description": "无手率最高阈值", + "key": "MAX_HAND_OCCLUSION", + "type": "string", + "value": "0.1" + } + ], + "id": "operator-builtin-video-filter", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-filter:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-filter@sha256:421e6b00389e9761b1bec6b316dec38151200f055ff9bd6deb6659ce170f97ae", + "inputs": [ + { + "description": "视频与元数据输入", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "数据过滤", + "outputs": [ + { + "description": "过滤结果", + "id": "out-1", + "name": "filtered", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_dataset_filter", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-video-filter", + "operatorSlug": "video_filter", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-artifacts" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_dataset_filter", + "runtimeSource": "backend/operators/builtin/video_filter/operator.py", + "skillDefinition": "针对 X-platform 质检结果执行真实筛选,输出符合条件的视频列表。", + "type": "builtin" + }, + { + "category": "VIDEO", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "将真实视频按固定时长切分为多个 mp4 片段。", + "hyperparams": [ + { + "description": "每个切片的秒数,显式覆盖时必须为正整数", + "key": "SEGMENT_DURATION_SECONDS", + "type": "string", + "value": "300" + } + ], + "id": "operator-builtin-video-slice", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-slice:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-slice@sha256:0102dbe48263fd4e6a62ab9ed48619d10a91c9b4f201ad0dc5b06071c62b05c1", + "inputs": [ + { + "description": "视频输入", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "视频切片", + "outputs": [ + { + "description": "切片结果", + "id": "out-1", + "name": "clips", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_video_slicing", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-video-slice", + "operatorSlug": "video_slice", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "ffmpeg", + "video-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_video_slicing", + "runtimeSource": "backend/operators/builtin/video_slice/operator.py", + "skillDefinition": "使用 ffmpeg/OpenCV 将真实视频切成固定长度片段,并输出 clip manifest。", + "type": "builtin" + }, + { + "category": "VIDEO", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "基于真实视频片段生成阶段划分、动作标签和对象提示。", + "hyperparams": [ + { + "description": "采样帧间隔秒数,显式覆盖时必须为大于 0 的有限数字", + "key": "FRAME_INTERVAL_SECONDS", + "type": "string", + "value": "1.0" + } + ], + "id": "operator-builtin-video-annotate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-annotate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-annotate@sha256:b2bb1476b0994f852a057c1b81bc412ed25c257f3581b720752c19435045f06d", + "inputs": [ + { + "description": "切片输入", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "视频标注", + "outputs": [ + { + "description": "标注结果", + "id": "out-1", + "name": "annotations", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_video_annotation", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-video-annotate", + "operatorSlug": "video_annotate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "video-files", + "json-artifacts" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_video_annotation", + "runtimeSource": "backend/operators/builtin/video_annotate/operator.py", + "skillDefinition": "读取切片视频并做真实帧间运动分析,输出 annotation JSON 和动作分段。", + "type": "builtin" + }, + { + "category": "VIDEO", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "对真实视频执行目标出现区间检测,并自动剪出命中片段。", + "hyperparams": [ + { + "description": "检测模式:motion、hog-person 或 opencv-dnn", + "key": "DETECTION_MODE", + "type": "string", + "value": "motion" + }, + { + "description": "目标标签,逗号分隔;motion 模式使用 motion,HOG 模式使用 person,DNN 模式使用模型类别名", + "key": "TARGET_LABELS", + "type": "string", + "value": "motion,person,object" + }, + { + "description": "采样帧间隔秒数,显式覆盖时必须为大于 0 的有限数字", + "key": "FRAME_INTERVAL_SECONDS", + "type": "string", + "value": "0.5" + }, + { + "description": "检测置信度阈值,范围 [0,1]", + "key": "MIN_CONFIDENCE", + "type": "string", + "value": "0.2" + }, + { + "description": "motion 模式最小运动面积占比,范围 [0,1]", + "key": "MIN_MOTION_AREA_RATIO", + "type": "string", + "value": "0.003" + }, + { + "description": "相邻命中段最大合并间隔秒数,显式覆盖时必须大于等于 0", + "key": "MERGE_GAP_SECONDS", + "type": "string", + "value": "0.5" + }, + { + "description": "输出片段最小时长秒数,显式覆盖时必须为大于 0 的有限数字", + "key": "MIN_SEGMENT_SECONDS", + "type": "string", + "value": "0.4" + }, + { + "description": "命中段前置缓冲秒数,显式覆盖时必须大于等于 0", + "key": "PRE_ROLL_SECONDS", + "type": "string", + "value": "0.1" + }, + { + "description": "命中段后置缓冲秒数,显式覆盖时必须大于等于 0", + "key": "POST_ROLL_SECONDS", + "type": "string", + "value": "0.2" + }, + { + "description": "最多输出目标片段数量,显式覆盖时必须为正整数", + "key": "MAX_CLIPS", + "type": "string", + "value": "20" + }, + { + "description": "opencv-dnn 模式的真实模型文件路径,例如 ONNX;motion/HOG 模式无需配置", + "key": "DNN_MODEL_PATH", + "type": "string", + "value": "" + }, + { + "description": "opencv-dnn 模式可选配置文件路径", + "key": "DNN_CONFIG_PATH", + "type": "string", + "value": "" + }, + { + "description": "opencv-dnn 模式可选类别名,逗号分隔;也可用 DNN_CLASS_NAMES_PATH 指向类别文件", + "key": "DNN_CLASS_NAMES", + "type": "string", + "value": "" + } + ], + "id": "operator-builtin-video-target-clip", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-target-clip:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-target-clip@sha256:0eaf4f0fbd20a6f017f1dd77a7d4660dc5cac1112dbbb32fc8e1a4c0cf6377f1", + "inputs": [ + { + "description": "视频清单或切片清单", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "视频目标检测裁剪", + "outputs": [ + { + "description": "目标出现片段清单", + "id": "out-1", + "name": "targetClips", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_video_target_clip", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-video-target-clip", + "operatorSlug": "video_target_clip", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "opencv-python-headless", + "numpy", + "ffmpeg", + "video-files", + "optional-opencv-dnn-model" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_video_target_clip", + "runtimeSource": "backend/operators/builtin/video_target_clip/operator.py", + "skillDefinition": "使用 OpenCV 对真实视频帧执行 motion / HOG person / OpenCV DNN 检测,按目标出现时间段剪辑 mp4 并输出 target clip manifest。", + "type": "builtin" + }, + { + "category": "VIDEO", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "将切片视频、质检结果、标注结果和配置文件组装成成品数据包。", + "id": "operator-builtin-package", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-package:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-package@sha256:e700d9d957364a1446c34f797fbdddd1c9e488c9e866913aeb740cb9b0a03741", + "inputs": [ + { + "description": "切片/标注/质检结果", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "成品打包", + "outputs": [ + { + "description": "成品包清单", + "id": "out-1", + "name": "packages", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_result_package", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-package", + "operatorSlug": "package", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "filesystem" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_result_package", + "runtimeSource": "backend/operators/builtin/package/operator.py", + "skillDefinition": "按 clip 组织真实 package 目录,输出 packages manifest 和所有成品文件。", + "type": "builtin" + }, + { + "category": "QA", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "按质检和标注规则对成品数据执行真实抽检采样。", + "hyperparams": [ + { + "description": "抽检规则数组;显式规则必须包含 id/field/operator/value/samplePercent,operator 必须受支持,samplePercent 必须在 (0,100]", + "key": "RULES_JSON", + "type": "string", + "value": "[]" + } + ], + "id": "operator-builtin-audit-sample", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-audit-sample:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-audit-sample@sha256:2a89baaae19e6c2c33980f41921020043888dc48db23cbc69c7de6f4670c9fc2", + "inputs": [ + { + "description": "成品包输入", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "抽检采样", + "outputs": [ + { + "description": "抽检样本清单", + "id": "out-1", + "name": "samples", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_audit_sampling", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-audit-sample", + "operatorSlug": "audit_sample", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-artifacts" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_audit_sampling", + "runtimeSource": "backend/operators/builtin/audit_sample/operator.py", + "skillDefinition": "根据 RULES_JSON 对成品数据包做规则筛选和稳定抽样,输出抽检任务清单。", + "type": "builtin" + }, + { + "category": "DATA", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "将上游 JSON 数据写入用户配置的业务 MongoDB,支持批量插入和按键 upsert,并允许显式覆盖连接。", + "hyperparams": [ + { + "description": "目标集合名称(禁止写系统保留集合)", + "key": "MONGO_COLLECTION", + "type": "string", + "value": "" + }, + { + "description": "写入模式:insert_many 或 upsert", + "key": "MONGO_MODE", + "type": "string", + "value": "insert_many" + }, + { + "description": "业务 MongoDB 连接 ID;推荐优先配置", + "key": "MONGO_CONNECTION_ID", + "type": "string", + "value": "" + }, + { + "description": "兼容字段;默认 business。若显式传入 URI/DB,会优先使用显式连接", + "key": "MONGO_TARGET", + "type": "string", + "value": "business" + }, + { + "description": "upsert 模式使用的唯一键", + "key": "UPSERT_KEY", + "type": "string", + "value": "_id" + }, + { + "description": "从输入 payload 中提取待写入文档的字段路径", + "key": "DOCUMENT_PATH", + "type": "string", + "value": "" + }, + { + "description": "可选数据库名;显式填写后会覆盖 business/platform 默认连接", + "key": "MONGO_DB", + "type": "string", + "value": "" + }, + { + "description": "可选 Mongo URI;显式填写后会覆盖 business/platform 默认连接", + "key": "MONGO_URI", + "type": "string", + "value": "" + } + ], + "id": "operator-builtin-mongodb-write", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-mongodb-write:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-mongodb-write@sha256:36677374caf4ee70072c0c6ec565498565d2b667c8fb238c9ebb81bf6ecee876", + "inputs": [ + { + "description": "待写入的 JSON 文档或 items 列表", + "id": "input", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "MongoDB 写入", + "outputs": [ + { + "description": "写入结果摘要", + "id": "result", + "name": "result", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_mongodb_write", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-mongodb-write", + "operatorSlug": "mongodb_write", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "mongodb" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_mongodb_write", + "runtimeSource": "backend/operators/builtin/mongodb_write/operator.py", + "skillDefinition": "将 JSON 文档写入业务 MongoDB 集合。", + "type": "builtin" + }, + { + "category": "DATASET", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "调用 Hugging Face Dataset Viewer API,读取真实数据集的 splits、size、parquet 分片和样本行。", + "hyperparams": [ + { + "description": "HF dataset repo id,例如 lerobot/pusht", + "key": "DATASET", + "type": "string", + "value": "" + }, + { + "description": "可选 config;为空时从 splits 自动选择 train 或首个 split", + "key": "CONFIG", + "type": "string", + "value": "" + }, + { + "description": "可选 split;为空时优先选择 train", + "key": "SPLIT", + "type": "string", + "value": "" + }, + { + "description": "样本行 offset,必须为非负整数", + "key": "ROW_OFFSET", + "type": "string", + "value": "0" + }, + { + "description": "样本行数量,范围 1-100", + "key": "ROW_LENGTH", + "type": "string", + "value": "20" + }, + { + "description": "是否调用 /rows 读取样本行", + "key": "INCLUDE_ROWS", + "type": "string", + "value": "true" + }, + { + "description": "是否调用 /parquet 读取 parquet 分片", + "key": "INCLUDE_PARQUET", + "type": "string", + "value": "true" + }, + { + "description": "是否调用 /statistics 读取列统计", + "key": "INCLUDE_STATISTICS", + "type": "string", + "value": "false" + }, + { + "description": "私有或 gated 数据集可填写 token;为空时不发送 Authorization", + "key": "HF_TOKEN", + "type": "string", + "value": "" + } + ], + "id": "operator-builtin-hf-dataset-probe", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-hf-dataset-probe:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-hf-dataset-probe@sha256:8ec0202d73059352be727b28c12daf9a6bdfd77e7e9ea3332d658efde66654c4", + "inputs": [ + { + "description": "可选上游数据集配置", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "HF 数据集探测", + "outputs": [ + { + "description": "HF 数据集探测结果", + "id": "out-1", + "name": "profile", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_hf_dataset_probe", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-hf-dataset-probe", + "operatorSlug": "hf_dataset_probe", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "huggingface-dataset-viewer-api" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_hf_dataset_probe", + "runtimeSource": "backend/operators/builtin/hf_dataset_probe/operator.py", + "skillDefinition": "面向 Hugging Face Hub/LeRobot 等公开或私有数据集,真实调用 Dataset Viewer API 获取可用 split、样本行、parquet 分片和规模信息;不生成模拟数据。", + "type": "builtin" + }, + { + "category": "DATASET", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 Parquet 文件或目录,汇总 schema、行数、row group、压缩编码和样本行,适配 Hugging Face/LeRobot 表格式数据入口。", + "hyperparams": [ + { + "description": "Parquet 文件或包含 .parquet 的目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "逗号分隔的必需列,例如 episode_index,frame_index,timestamp,action", + "key": "REQUIRED_COLUMNS", + "type": "string", + "value": "" + }, + { + "description": "逗号分隔的样本列;为空时读取首批全部列", + "key": "SAMPLE_COLUMNS", + "type": "string", + "value": "" + }, + { + "description": "最多扫描 Parquet 文件数", + "key": "MAX_FILES", + "type": "string", + "value": "50" + }, + { + "description": "每个文件最多读取的样本行数;0 表示不读取样本行", + "key": "MAX_SAMPLE_ROWS", + "type": "string", + "value": "5" + } + ], + "id": "operator-builtin-parquet-table-summary", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-parquet-table-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-parquet-table-summary@sha256:03e9c35aa6e3658bddd4821d6125b5bff10b4d412d33d655dfd869cdae3d03fb", + "inputs": [ + { + "description": "Parquet 文件、目录或上游 artifact", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "Parquet 表摘要", + "outputs": [ + { + "description": "Parquet 表结构和样本摘要", + "id": "out-1", + "name": "summary", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_parquet_table_summary", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-parquet-table-summary", + "operatorSlug": "parquet_table_summary", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "parquet-files", + "pyarrow" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_parquet_table_summary", + "runtimeSource": "backend/operators/builtin/parquet_table_summary/operator.py", + "skillDefinition": "面向 Hugging Face Dataset、LeRobot v3、轨迹元数据和 episode/action 表,真实读取 .parquet metadata 与样本行,校验必需列是否存在,不通过固定字段伪造输出。", + "type": "builtin" + }, + { + "category": "DATASET", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 COCO annotations JSON,校验 image/category 引用、bbox 边界、图片文件存在性和类别分布。", + "hyperparams": [ + { + "description": "COCO annotation JSON 或数据集根目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "可选图片根目录;为空时从 images/ 和 annotation 相邻目录推断", + "key": "IMAGE_ROOT", + "type": "string", + "value": "" + }, + { + "description": "每类问题最多保留样例数", + "key": "MAX_ISSUES", + "type": "string", + "value": "100" + } + ], + "id": "operator-builtin-coco-annotation-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-coco-annotation-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-coco-annotation-validate@sha256:538e5ffdc881f6f93ae55b84c216b57954bbf662a1d1a34588ed4981eb8ab57d", + "inputs": [ + { + "description": "COCO annotations JSON 或数据集根目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "COCO 标注校验", + "outputs": [ + { + "description": "COCO 标注校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_coco_annotation_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-coco-annotation-validate", + "operatorSlug": "coco_annotation_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "coco-json", + "image-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_coco_annotation_validate", + "runtimeSource": "backend/operators/builtin/coco_annotation_validate/operator.py", + "skillDefinition": "面向机器人感知、抓取检测、实例分割等 COCO 风格标注集,真实解析 images/annotations/categories,检查 bbox 是否越界、类别和图片引用是否存在,并统计 segmentation 覆盖。", + "type": "builtin" + }, + { + "category": "DATASET", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 YOLO labels txt、images 目录和 data.yaml/classes,校验归一化 bbox、类别 ID、图片/标签配对和类别分布。", + "hyperparams": [ + { + "description": "YOLO 数据集根目录或 labels 文件", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "逗号分隔类别名;为空时读取 data.yaml/dataset.yaml 的 names", + "key": "CLASS_NAMES", + "type": "string", + "value": "" + }, + { + "description": "可选图片根目录;为空时从 SOURCE_PATH 推断", + "key": "IMAGE_ROOT", + "type": "string", + "value": "" + }, + { + "description": "可选标签根目录;为空时从 SOURCE_PATH 推断", + "key": "LABEL_ROOT", + "type": "string", + "value": "" + }, + { + "description": "每类问题最多保留样例数", + "key": "MAX_ISSUES", + "type": "string", + "value": "100" + } + ], + "id": "operator-builtin-yolo-detection-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-yolo-detection-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-yolo-detection-validate@sha256:2478bdf9dd3df22a9d5d6306c8ad0180891efc3d82bd028b50e5abfbc0e7783d", + "inputs": [ + { + "description": "YOLO 数据集根目录或 labels 文件", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "YOLO 检测标注校验", + "outputs": [ + { + "description": "YOLO 检测标注校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_yolo_detection_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-yolo-detection-validate", + "operatorSlug": "yolo_detection_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "yolo-label-files", + "image-files", + "pyyaml" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_yolo_detection_validate", + "runtimeSource": "backend/operators/builtin/yolo_detection_validate/operator.py", + "skillDefinition": "面向机器人目标检测、抓取区域识别和合成视觉数据训练,真实读取 YOLO txt 标注、data.yaml 类别和图片目录,检查 class id、归一化 bbox 范围、图片/标签成对关系;不接受空跑通过。", + "type": "builtin" + }, + { + "category": "DATASET", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "基于真实 episode/demo 清单按稳定哈希生成 train/val/test 切分,并输出 JSON 与 CSV 清单。", + "hyperparams": [ + { + "description": "可选 episode JSON/JSONL、LeRobot 根目录或包含 episodes.jsonl 的目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "训练集比例,必须为非负数字", + "key": "TRAIN_RATIO", + "type": "string", + "value": "0.8" + }, + { + "description": "验证集比例,必须为非负数字", + "key": "VAL_RATIO", + "type": "string", + "value": "0.1" + }, + { + "description": "测试集比例,必须为非负数字", + "key": "TEST_RATIO", + "type": "string", + "value": "0.1" + }, + { + "description": "稳定哈希种子,同一数据和 seed 会生成同一切分", + "key": "SEED", + "type": "string", + "value": "droboticflow" + }, + { + "description": "低于该 step 数的 episode/demo 会被跳过", + "key": "MIN_STEPS", + "type": "string", + "value": "1" + } + ], + "id": "operator-builtin-dataset-split-manifest", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dataset-split-manifest:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dataset-split-manifest@sha256:b2bcd44ccaa4b3c2b9a22d4f10a5d8101a0e19e249e693a501e1e506887fe1da", + "inputs": [ + { + "description": "包含 episodes/demos/items 的上游摘要或真实清单", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "训练切分清单生成", + "outputs": [ + { + "description": "训练切分 JSON 清单", + "id": "out-1", + "name": "manifest", + "transmission": "bus", + "type": "json" + }, + { + "description": "训练切分 CSV 清单", + "id": "out-2", + "name": "csv", + "transmission": "bus", + "type": "file" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_dataset_split_manifest", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-dataset-split-manifest", + "operatorSlug": "dataset_split_manifest", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-artifacts", + "csv-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_dataset_split_manifest", + "runtimeSource": "backend/operators/builtin/dataset_split_manifest/operator.py", + "skillDefinition": "面向 LeRobot、RLDS、HDF5/robomimic 等具身数据训练入口,读取真实 episode/demo metadata,用稳定哈希生成可复现的 train/val/test 清单;不按硬编码 ID 切分。", + "type": "builtin" + }, + { + "category": "DATASET", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实数据集目录中的 Dataset Card、LICENSE、CITATION 和 provenance metadata,输出来源许可审计结果。", + "hyperparams": [ + { + "description": "包含 README.md、LICENSE、CITATION 或 provenance.json 的真实数据集目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + } + ], + "id": "operator-builtin-dataset-provenance-audit", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dataset-provenance-audit:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dataset-provenance-audit@sha256:641ef1697d6e61773dc587c9142a50b920738107974f7ffbacf2512302aee98f", + "inputs": [ + { + "description": "数据集目录或上游目录 artifact", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "数据集来源许可审计", + "outputs": [ + { + "description": "数据集来源许可审计结果", + "id": "out-1", + "name": "audit", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_dataset_provenance_audit", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-dataset-provenance-audit", + "operatorSlug": "dataset_provenance_audit", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "dataset-card-files", + "license-files", + "pyyaml" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_dataset_provenance_audit", + "runtimeSource": "backend/operators/builtin/dataset_provenance_audit/operator.py", + "skillDefinition": "面向 Hugging Face Dataset Card、GitHub 数据仓库和内部数据集发布前合规检查,真实读取 README front matter、LICENSE、CITATION、provenance.json 等文件,识别缺失 license/source/citation 的数据集。", + "type": "builtin" + }, + { + "category": "DATASET", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 WebDataset tar/tar.gz 分片,统计样本键、扩展名、模态组合和不安全成员路径。", + "hyperparams": [ + { + "description": ".tar/.tar.gz 文件或包含 WebDataset 分片的目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多扫描多少个 shard,必须为正整数", + "key": "MAX_SHARDS", + "type": "string", + "value": "20" + }, + { + "description": "每个 shard 最多返回多少个成员和样本预览", + "key": "MAX_PREVIEW_ENTRIES", + "type": "string", + "value": "20" + } + ], + "id": "operator-builtin-webdataset-shard-summary", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-webdataset-shard-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-webdataset-shard-summary@sha256:ce71042733709b064da7e6ceeadfa4a3f4c138d627da5ff9fe7b877d1de2b375", + "inputs": [ + { + "description": "WebDataset tar shard 文件或目录输入", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "WebDataset 分片概览", + "outputs": [ + { + "description": "WebDataset shard 样本/模态概览", + "id": "out-1", + "name": "summary", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_webdataset_shard_summary", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-webdataset-shard-summary", + "operatorSlug": "webdataset_shard_summary", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "webdataset-tar-shards", + "python-tarfile" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_webdataset_shard_summary", + "runtimeSource": "backend/operators/builtin/webdataset_shard_summary/operator.py", + "skillDefinition": "面向 Hugging Face WebDataset、LAION 风格 tar shard 和内部大规模感知数据,真实读取 tar 成员并按同名 sample key 聚合图片、JSON、caption、tensor 等模态,不靠文件名硬编码结果。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 AgiBotWorld2026 meta/info、episode 元数据、data Parquet 和视频路径模板,校验 episode 长度、必需列、任务索引和视频文件可读性。", + "hyperparams": [ + { + "description": "AgiBotWorld2026 数据集根目录或其中任意文件路径", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多扫描多少个 episode/data Parquet 文件", + "key": "MAX_PARQUET_FILES", + "type": "string", + "value": "100" + }, + { + "description": "未显式提供 chunk_index 时用于 episode_index 推导 chunk 的大小", + "key": "CHUNK_SIZE", + "type": "string", + "value": "1000" + }, + { + "description": "data Parquet 必须包含的列,逗号分隔", + "key": "REQUIRED_COLUMNS", + "type": "string", + "value": "episode_index,frame_index,timestamp,action" + }, + { + "description": "可选 data path 模板,例如 data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", + "key": "DATA_PATH_TEMPLATE", + "type": "string", + "value": "" + }, + { + "description": "可选 video path 模板,例如 videos/{video_key}/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.mp4", + "key": "VIDEO_PATH_TEMPLATE", + "type": "string", + "value": "" + }, + { + "description": "可选视频 feature key 列表;为空时从 meta/info.json features 中 dtype=video 的字段推导", + "key": "VIDEO_KEYS", + "type": "string", + "value": "" + }, + { + "description": "是否要求 video_path 模板对应的视频文件真实存在", + "key": "CHECK_VIDEO_EXISTS", + "type": "string", + "value": "true" + } + ], + "id": "operator-builtin-agibot-world2026-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-agibot-world2026-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-agibot-world2026-validate@sha256:622499759d3ae4bb0bbcbd51069fb2525510f41d3a42a474e30b7d21d9df41b8", + "inputs": [ + { + "description": "AgiBotWorld2026 数据集根目录或上游目录 artifact", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "AgiBotWorld2026 数据集校验", + "outputs": [ + { + "description": "AgiBotWorld2026 parquet/video episode 结构校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_agibot_world2026_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-agibot-world2026-validate", + "operatorSlug": "agibot_world2026_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "agibotworld2026-meta-data-video-files", + "pyarrow", + "parquet-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_agibot_world2026_validate", + "runtimeSource": "backend/operators/builtin/agibot_world2026_validate/operator.py", + "skillDefinition": "面向 Hugging Face/ModelScope 上的 AgiBotWorld2026 大规模具身数据格式,真实读取 meta/info.json、meta/episodes.jsonl 或 meta/episodes/*.parquet、data/**/*.parquet,以及 video_path 模板指向的相机流文件,补齐训练前对 episode 行数、action/timestamp/frame_index 列、task 引用和视频缺失的验收。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 RLDS/TFDS 导出的 episodes/steps JSON 或目录元数据,输出轨迹长度、动作维度、观测键和终止标志校验。", + "hyperparams": [ + { + "description": "RLDS JSON/JSONL 文件、导出目录或 TFDS 元数据目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + } + ], + "id": "operator-builtin-rlds-episode-summary", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rlds-episode-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rlds-episode-summary@sha256:49aae340f8d0d11ebdf89f9a01946477472c9b09b34e7f51fdef51c2feffabcb", + "inputs": [ + { + "description": "RLDS JSON/JSONL 或目录输入", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "RLDS 轨迹概览", + "outputs": [ + { + "description": "RLDS 轨迹概览", + "id": "out-1", + "name": "summary", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_rlds_episode_summary", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-rlds-episode-summary", + "operatorSlug": "rlds_episode_summary", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "rlds-json-or-tfds-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_rlds_episode_summary", + "runtimeSource": "backend/operators/builtin/rlds_episode_summary/operator.py", + "skillDefinition": "面向 Open X-Embodiment/RLDS 风格 trajectory 数据,读取真实 episode/step 结构并校验 is_first/is_last/is_terminal、reward、action、observation 等关键字段。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 BridgeData/TFDS-RLDS 目录,校验 dataset_info/features 中的 image/action/language 特征声明,并逐条扫描 TFRecord shard framing 与 CRC。", + "hyperparams": [ + { + "description": "BridgeData/TFDS-RLDS 数据集目录或 TFRecord 文件", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多扫描多少个 TFRecord shard", + "key": "MAX_FILES", + "type": "string", + "value": "100" + }, + { + "description": "每个 shard 最多读取多少条 record", + "key": "MAX_RECORDS", + "type": "string", + "value": "200000" + }, + { + "description": "每个 shard 最多返回多少条 payload 预览", + "key": "PREVIEW_RECORDS", + "type": "string", + "value": "3" + }, + { + "description": "特征清单中用于判定图像观测字段的 token 列表", + "key": "IMAGE_FIELD_TOKENS", + "type": "string", + "value": "image,observation.image,observation.images" + }, + { + "description": "特征清单中用于判定动作字段的 token 列表", + "key": "ACTION_FIELD_TOKENS", + "type": "string", + "value": "action" + }, + { + "description": "特征清单中用于判定语言指令字段的 token 列表", + "key": "LANGUAGE_FIELD_TOKENS", + "type": "string", + "value": "language_instruction,natural_language_instruction,instruction" + } + ], + "id": "operator-builtin-bridgedata-tfds-manifest-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-bridgedata-tfds-manifest-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-bridgedata-tfds-manifest-validate@sha256:18c89efc04d3ca65154c07851231c16953900016178d6183041850e6e7f7e2a0", + "inputs": [ + { + "description": "BridgeData/TFDS-RLDS 数据集目录、TFDS 导出目录或 TFRecord 文件", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "BridgeData TFDS-RLDS清单校验", + "outputs": [ + { + "description": "BridgeData TFDS-RLDS 清单校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_bridgedata_tfds_manifest_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-bridgedata-tfds-manifest-validate", + "operatorSlug": "bridgedata_tfds_manifest_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "bridgedata-tfds-rlds-directory", + "tfrecord-files", + "crc32c-structural-validation" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_bridgedata_tfds_manifest_validate", + "runtimeSource": "backend/operators/builtin/bridgedata_tfds_manifest_validate/operator.py", + "skillDefinition": "面向 BridgeData V2、Open X-Embodiment 和 TFDS/RLDS 分片数据,真实读取 dataset_info.json、features.json 和 .tfrecord/.tfrec shard,校验 image/action/language_instruction 等训练关键字段声明、split record 数量、TFRecord length/header/payload CRC,避免只验证文件存在而漏掉损坏 shard 或缺少语言/动作特征。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 DROID raw episode 目录,校验 trajectory.h5、metadata_*.json 和 recordings/MP4 视频文件,并检查 action/state 数据集帧数一致性。", + "hyperparams": [ + { + "description": "DROID raw episode 目录或包含多个 episode 的根目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多校验多少个 episode 目录,必须为正整数", + "key": "MAX_EPISODES", + "type": "string", + "value": "100" + }, + { + "description": "每个 HDF5 dataset 最多抽样多少行用于有限值检查", + "key": "MAX_HDF5_ROWS", + "type": "string", + "value": "200000" + }, + { + "description": "是否要求 HDF5 中存在 action/control 相关数据集", + "key": "REQUIRE_ACTION", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求 HDF5 中存在 state/proprio/joint/gripper 相关数据集", + "key": "REQUIRE_STATE", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求 episode 中存在 recordings/MP4 下的 MP4 视频文件", + "key": "CHECK_VIDEO_EXISTS", + "type": "boolean", + "value": "true" + } + ], + "id": "operator-builtin-droid-raw-episode-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-droid-raw-episode-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-droid-raw-episode-validate@sha256:eea450070b6b1fa4fbd80c9f1e476eacba25683791b0b40320639c65cc5fa418", + "inputs": [ + { + "description": "DROID raw episode 目录或包含多个 episode 的根目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "DROID Raw Episode校验", + "outputs": [ + { + "description": "DROID raw episode 文件结构和轨迹数据校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_droid_raw_episode_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-droid-raw-episode-validate", + "operatorSlug": "droid_raw_episode_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "droid-raw-episode-hdf5-metadata-video-files", + "h5py", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_droid_raw_episode_validate", + "runtimeSource": "backend/operators/builtin/droid_raw_episode_validate/operator.py", + "skillDefinition": "面向 DROID raw dataset 和 GitHub/HuggingFace 常见 DROID episode 导出,真实读取 trajectory.h5/trajectory_*.h5 HDF5 数据集、metadata_*.json 元数据和 recordings/MP4/*.mp4 视频,校验动作数据集、状态/关节/夹爪数据集、帧数一致性和视频文件存在性。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 LIBERO HDF5 demonstration 文件,校验 /data/demo_*、actions、states、obs 图像/proprio 字段、语言任务字段和 BDDL 任务文件匹配关系。", + "hyperparams": [ + { + "description": "LIBERO .hdf5/.h5 文件或包含多个 task demo 的目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "可选 BDDL 文件或目录;为空时会在 SOURCE_PATH 附近自动查找 .bddl", + "key": "BDDL_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多校验多少个 HDF5 task 文件", + "key": "MAX_FILES", + "type": "string", + "value": "100" + }, + { + "description": "每个文件最多读取多少个 demo group", + "key": "MAX_DEMOS", + "type": "string", + "value": "1000" + }, + { + "description": "每个 dataset 最多抽样多少行用于有限值检查", + "key": "MAX_ROWS", + "type": "string", + "value": "200000" + }, + { + "description": "必须存在的 obs 数据集路径,逗号分隔", + "key": "REQUIRED_OBS_DATASETS", + "type": "string", + "value": "agentview_rgb,eye_in_hand_rgb,gripper_states,joint_states,ee_states" + }, + { + "description": "是否要求存在与 HDF5 task 同名的 .bddl 文件", + "key": "REQUIRE_BDDL", + "type": "boolean", + "value": "false" + }, + { + "description": "是否要求每个 demo 包含 states 数据集", + "key": "REQUIRE_STATES", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求 HDF5 attrs 或 dataset 中存在 language/instruction/task 字段", + "key": "REQUIRE_LANGUAGE", + "type": "boolean", + "value": "false" + }, + { + "description": "是否在存在 dones 数据集时要求最后一帧 done=true", + "key": "REQUIRE_TERMINAL_DONE", + "type": "boolean", + "value": "false" + } + ], + "id": "operator-builtin-libero-hdf5-task-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-libero-hdf5-task-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-libero-hdf5-task-validate@sha256:68b8f8d5b203a0c644d8cec121c6d802ec55cf89ac0fa4aa87eb471b7076e18d", + "inputs": [ + { + "description": "LIBERO .hdf5/.h5 文件或包含多个 task demo 的目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "LIBERO HDF5任务校验", + "outputs": [ + { + "description": "LIBERO HDF5 task demonstration 结构校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_libero_hdf5_task_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-libero-hdf5-task-validate", + "operatorSlug": "libero_hdf5_task_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "libero-hdf5-demo-files", + "libero-bddl-files", + "h5py", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_libero_hdf5_task_validate", + "runtimeSource": "backend/operators/builtin/libero_hdf5_task_validate/operator.py", + "skillDefinition": "面向 LIBERO、OpenVLA LIBERO fine-tuning 和 GitHub/HuggingFace 常见 LIBERO 数据集导出,真实读取 *_demo.hdf5 中 /data/demo_* 的 actions、states、obs/agentview_rgb、obs/eye_in_hand_rgb、obs/gripper_states、obs/joint_states、obs/ee_states、dones/rewards 和语言字段,并可核对同名 .bddl 任务定义文件。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 RoboCasa HDF5 demonstration 文件,校验 data/demo_*、actions、action_dict、states、obs 图像/proprio 字段、ep_meta 语言任务和 model_file 回放元数据。", + "hyperparams": [ + { + "description": "RoboCasa .hdf5/.h5 文件或包含多个 demonstration 文件的目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多校验多少个 HDF5 文件", + "key": "MAX_FILES", + "type": "string", + "value": "100" + }, + { + "description": "每个文件最多读取多少个 demo group", + "key": "MAX_DEMOS", + "type": "string", + "value": "1000" + }, + { + "description": "每个 dataset 最多抽样多少行用于有限值检查", + "key": "MAX_ROWS", + "type": "string", + "value": "200000" + }, + { + "description": "用于识别图像观测字段的 token 列表", + "key": "IMAGE_OBS_TOKENS", + "type": "string", + "value": "image,rgb,depth,agentview,eye_in_hand" + }, + { + "description": "用于识别 proprio/机器人状态观测字段的 token 列表", + "key": "PROPRIO_OBS_TOKENS", + "type": "string", + "value": "state,joint,gripper,eef,ee_,proprio,qpos,qvel" + }, + { + "description": "是否要求 /data attrs 中存在可解析的 env_args", + "key": "REQUIRE_ENV_ARGS", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求每个 demo 有 model_file 回放模型元数据", + "key": "REQUIRE_MODEL_FILE", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求每个 demo 有可解析的 ep_meta", + "key": "REQUIRE_EP_META", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求每个 demo 包含 action_dict 分量数据", + "key": "REQUIRE_ACTION_DICT", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求每个 demo 包含 states 回放状态", + "key": "REQUIRE_STATES", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求 ep_meta 中存在 lang/language/task/instruction 字段", + "key": "REQUIRE_LANGUAGE", + "type": "boolean", + "value": "true" + } + ], + "id": "operator-builtin-robocasa-hdf5-demonstration-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-robocasa-hdf5-demonstration-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-robocasa-hdf5-demonstration-validate@sha256:d6010ee3a8123e3b67d406742ac822911ea556f69320ef5603edba24a3b691d5", + "inputs": [ + { + "description": "RoboCasa .hdf5/.h5 文件或包含多个 demonstration 文件的目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "RoboCasa HDF5示教校验", + "outputs": [ + { + "description": "RoboCasa HDF5 demonstration 结构校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_robocasa_hdf5_demonstration_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-robocasa-hdf5-demonstration-validate", + "operatorSlug": "robocasa_hdf5_demonstration_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "robocasa-hdf5-demo-files", + "h5py", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_robocasa_hdf5_demonstration_validate", + "runtimeSource": "backend/operators/builtin/robocasa_hdf5_demonstration_validate/operator.py", + "skillDefinition": "面向 RoboCasa、robomimic/RoboSuite 派生家庭任务示教和 GitHub/HuggingFace 常见 RoboCasa 原始 HDF5 数据,真实读取 /data/demo_* 下 actions、action_dict、states、obs/next_obs,多相机图像、关节/夹爪/eef proprio 字段,并校验 demo.attrs.ep_meta、demo.attrs.model_file 和 data.attrs.env_args。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 FurnitureBench .pkl 示教文件,校验 observations、两路彩色图、robot_state、8 维 actions、rewards、skills 和 low/med/high 目录布局。", + "hyperparams": [ + { + "description": "FurnitureBench .pkl 文件、家具目录或包含 low/med/high 的数据集根目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多校验多少个 .pkl/.pickle 示教文件,必须为正整数", + "key": "MAX_FILES", + "type": "string", + "value": "100" + }, + { + "description": "期望 action 维度;FurnitureBench 官方格式为 8-D actions", + "key": "EXPECTED_ACTION_DIM", + "type": "string", + "value": "8" + }, + { + "description": "每帧 observation 必须包含的图像字段,默认 wrist/front 两路彩色相机", + "key": "REQUIRED_IMAGE_KEYS", + "type": "string", + "value": "color_image1,color_image2" + }, + { + "description": "每帧 robot_state 必须包含的状态字段", + "key": "REQUIRED_ROBOT_STATE_KEYS", + "type": "string", + "value": "ee_pos,ee_quat,ee_pos_vel,ee_ori_vel,joint_positions,joint_velocities,joint_torques,gripper_width" + }, + { + "description": "是否要求 skills 列表存在且长度与 observations 一致", + "key": "REQUIRE_SKILLS", + "type": "boolean", + "value": "true" + } + ], + "id": "operator-builtin-furniturebench-pkl-demonstration-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-furniturebench-pkl-demonstration-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-furniturebench-pkl-demonstration-validate@sha256:6633bd68e09a6a82997e46bd6f02a0a3b27f77ba40aab0094239cd11317f7c8c", + "inputs": [ + { + "description": "FurnitureBench .pkl 文件或 low/med/high 数据集根目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "FurnitureBench PKL示教校验", + "outputs": [ + { + "description": "FurnitureBench PKL 示教结构校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_furniturebench_pkl_demonstration_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-furniturebench-pkl-demonstration-validate", + "operatorSlug": "furniturebench_pkl_demonstration_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "furniturebench-pkl-files", + "pickle", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_furniturebench_pkl_demonstration_validate", + "runtimeSource": "backend/operators/builtin/furniturebench_pkl_demonstration_validate/operator.py", + "skillDefinition": "面向 FurnitureBench 真实家具装配示教数据,读取 low/med/high//.pkl 或单个 .pkl 文件,校验 furniture、observations 列表、color_image1/color_image2、robot_state 中 ee_pos/ee_quat/关节/夹爪等状态、8-D actions、rewards 和 skills 的长度一致性、shape 与有限数值,适合在训练前发现损坏 pickle、缺失相机/状态字段和动作维度漂移。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取 RoboTurk Real Robot Dataset aligned HDF5,校验 data/user/demo 层级、视频路径属性、robot_observation ROS 观测和 user_control 控制序列。", + "hyperparams": [ + { + "description": "RoboTurk *_aligned_dataset.hdf5 文件或数据集目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多校验多少个 .hdf5/.h5 文件,必须为正整数", + "key": "MAX_FILES", + "type": "string", + "value": "20" + }, + { + "description": "每个 HDF5 文件最多展开多少个 demo 明细,必须为正整数", + "key": "MAX_DEMOS_PER_FILE", + "type": "string", + "value": "50" + }, + { + "description": "每个数值 dataset 最多抽样多少行做 finite 检查", + "key": "MAX_NUMERIC_SAMPLE_ROWS", + "type": "string", + "value": "256" + }, + { + "description": "robot_observation 中必须存在的数据集字段", + "key": "REQUIRED_ROBOT_OBSERVATION_DATASETS", + "type": "string", + "value": "eef_poses,eef_times,joint_states_arm,joint_states_arm_times,joint_states_gripper,joint_states_gripper_times" + }, + { + "description": "user_control 中必须存在的数据集字段", + "key": "REQUIRED_USER_CONTROL_DATASETS", + "type": "string", + "value": "user_control,user_control_times" + }, + { + "description": "是否要求 top_depth_video_file、top_rgb_video_file、front_rgb_video_file 视频路径属性", + "key": "REQUIRE_VIDEO_ATTRS", + "type": "boolean", + "value": "true" + }, + { + "description": "是否抽样检查数值 dataset 不含 NaN/Inf", + "key": "REQUIRE_FINITE_NUMERIC", + "type": "boolean", + "value": "true" + } + ], + "id": "operator-builtin-roboturk-real-hdf5-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-roboturk-real-hdf5-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-roboturk-real-hdf5-validate@sha256:db43d615bf52f8ea734d8a45e001602e91b32c7b2046e51cade16489faf4c838", + "inputs": [ + { + "description": "RoboTurk *_aligned_dataset.hdf5 文件或包含该文件的数据集目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "RoboTurk Real HDF5校验", + "outputs": [ + { + "description": "RoboTurk Real HDF5 结构与数值校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_roboturk_real_hdf5_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-roboturk-real-hdf5-validate", + "operatorSlug": "roboturk_real_hdf5_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "roboturk-real-aligned-hdf5-files", + "h5py", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_roboturk_real_hdf5_validate", + "runtimeSource": "backend/operators/builtin/roboturk_real_hdf5_validate/operator.py", + "skillDefinition": "面向 RoboTurk Real Robot Dataset 的真实 aligned HDF5,读取 data//demo_* 层级,校验 total_demos/total_time、clock_time/session_id/demo_time/date/task、Kinect/USB 视频路径属性、robot_observation 中 eef_poses/eef_times/joint_states_arm/joint_states_arm_times/joint_states_gripper/joint_states_gripper_times,以及 user_control/user_control_times 的 shape、时间戳长度一致性和有限数值,适合在视频预测、模仿学习和真实机器人遥操作数据入库前发现结构漂移。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取 RoboSet/RoboHive .h5/.hdf5 轨迹文件,校验 Trial*/data/derived/config 层级、actions、rewards、done/success、observation/state 和多相机 RGB 数据。", + "hyperparams": [ + { + "description": "RoboSet .h5/.hdf5 文件或数据集目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多校验多少个 .h5/.hdf5 文件,必须为正整数", + "key": "MAX_FILES", + "type": "string", + "value": "20" + }, + { + "description": "每个 HDF5 文件最多展开多少个 Trial 明细,必须为正整数", + "key": "MAX_TRIALS_PER_FILE", + "type": "string", + "value": "50" + }, + { + "description": "每个数值 dataset 最多抽样多少行做 finite 检查", + "key": "MAX_NUMERIC_SAMPLE_ROWS", + "type": "string", + "value": "256" + }, + { + "description": "每个 Trial group 必须存在的一级 key", + "key": "REQUIRED_TRIAL_KEYS", + "type": "string", + "value": "data,derived,config" + }, + { + "description": "识别 action 序列时使用的数据集候选名", + "key": "ACTION_KEY_CANDIDATES", + "type": "string", + "value": "action,actions,u,ctrl,control" + }, + { + "description": "识别 reward 序列时使用的数据集候选名", + "key": "REWARD_KEY_CANDIDATES", + "type": "string", + "value": "reward,rewards" + }, + { + "description": "识别终止或成功标志序列时使用的数据集候选名", + "key": "DONE_KEY_CANDIDATES", + "type": "string", + "value": "done,dones,terminal,terminals,success,successes" + }, + { + "description": "识别 observation/state 序列时使用的数据集候选名", + "key": "OBSERVATION_KEY_CANDIDATES", + "type": "string", + "value": "obs,observation,observations,state,states,qpos,qvel,proprio,proprioception" + }, + { + "description": "识别多相机 RGB/image 数据集时使用的路径 token", + "key": "RGB_KEY_TOKENS", + "type": "string", + "value": "rgb,image,images,camera,cam" + }, + { + "description": "每个 Trial 至少需要识别到多少路 RGB/image 数据集;真实 kitchen 数据可设为 4", + "key": "MIN_RGB_VIEWS", + "type": "string", + "value": "1" + }, + { + "description": "是否要求每个 Trial 包含 action 序列", + "key": "REQUIRE_ACTIONS", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求每个 Trial 包含 reward 序列", + "key": "REQUIRE_REWARDS", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求每个 Trial 包含 RGB/image 视觉序列", + "key": "REQUIRE_RGB", + "type": "boolean", + "value": "true" + }, + { + "description": "是否抽样检查数值 dataset 不含 NaN/Inf", + "key": "REQUIRE_FINITE_NUMERIC", + "type": "boolean", + "value": "true" + } + ], + "id": "operator-builtin-roboset-hdf5-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-roboset-hdf5-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-roboset-hdf5-validate@sha256:db36835317e9ce8c8fa5d550ddf8327460f2a891bf57a2219bd874e6eb154ec1", + "inputs": [ + { + "description": "RoboSet .h5/.hdf5 文件或包含多个轨迹文件的数据集目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "RoboSet HDF5轨迹校验", + "outputs": [ + { + "description": "RoboSet HDF5 轨迹结构与数值校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_roboset_hdf5_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-roboset-hdf5-validate", + "operatorSlug": "roboset_hdf5_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "roboset-robohive-hdf5-trajectory-files", + "h5py", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_roboset_hdf5_validate", + "runtimeSource": "backend/operators/builtin/roboset_hdf5_validate/operator.py", + "skillDefinition": "面向 RoboSet/RoboHive 真实 HDF5 轨迹数据,读取 Trial0/Trial1/... 下的 data、derived、config,递归扫描 data 中 actions、rewards、done/success、observations/state/qpos/qvel、rgb/camera/image 多相机视觉数组,校验关键 group 存在、帧长一致性、RGB 视图数量、action 维度分布和有限数值,适合真实厨房遥操作、kinesthetic playback 与 RoboHive 数据入库前做格式漂移和损坏文件排查。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 RH20T scene 目录,校验 metadata.json、cam_* 多相机媒体与 timestamps.npy、transformed 力/力矩 TCP 关节夹爪数组、高频数据和 calibration 标定文件。", + "hyperparams": [ + { + "description": "RH20T scene 目录或包含多个 scene 的根目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多校验多少个 scene 目录", + "key": "MAX_SCENES", + "type": "string", + "value": "100" + }, + { + "description": "transformed 目录下必须存在的 .npy 文件,逗号分隔", + "key": "REQUIRED_TRANSFORMED_FILES", + "type": "string", + "value": "force_torque.npy,tcp.npy,joint.npy,gripper.npy" + }, + { + "description": "是否要求 scene 下存在可解析 metadata.json", + "key": "REQUIRE_METADATA", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求 scene 下存在 cam_* 相机目录", + "key": "REQUIRE_CAMERAS", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求 transformed 目录及关键力/位姿/关节/夹爪数组", + "key": "REQUIRE_TRANSFORMED", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求 metadata calib 指向的 intrinsics/extrinsics/tcp 标定文件", + "key": "REQUIRE_CALIBRATION", + "type": "boolean", + "value": "true" + } + ], + "id": "operator-builtin-rh20t-scene-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rh20t-scene-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rh20t-scene-validate@sha256:ba1e0190c4d8283480bbed515f6a3ba26132bfc4fb74ec701b137b043f194c94", + "inputs": [ + { + "description": "RH20T scene 目录或包含多个 scene 的根目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "RH20T 场景结构校验", + "outputs": [ + { + "description": "RH20T scene 结构和关键数据文件校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_rh20t_scene_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-rh20t-scene-validate", + "operatorSlug": "rh20t_scene_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "rh20t-scene-directory", + "numpy", + "npy-files", + "image-video-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_rh20t_scene_validate", + "runtimeSource": "backend/operators/builtin/rh20t_scene_validate/operator.py", + "skillDefinition": "面向 RH20T、Real Human 20T、真实遥操作多模态机器人数据,读取 scene folder 中 cam_* 相机目录、timestamps.npy、RGB/视频媒体、metadata.json、transformed/force_torque.npy、tcp.npy、joint.npy、gripper.npy、high_freq_data 和 calib 标定 npy 文件,检查目录结构、关键模态存在、时间戳单调性、数组 shape/finite 数值和 calibration 完整性。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 RoboMIND trajectory.hdf5,校验 puppet/master joint_position、puppet/end_effector 卡死、非有限值、帧数和多模态字段。", + "hyperparams": [ + { + "description": "RoboMIND trajectory.hdf5 文件、data 目录或包含 benchmark/h5_* 的根目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多校验多少个 .hdf5/.h5 轨迹文件,优先读取 trajectory.hdf5", + "key": "MAX_FILES", + "type": "string", + "value": "100" + }, + { + "description": "每个数值数组最多采样多少帧用于 range 和 unique ratio 计算", + "key": "MAX_ROWS", + "type": "string", + "value": "200000" + }, + { + "description": "末端执行器数据 max(range_per_dim) 小于该阈值时判定卡死", + "key": "EE_TOLERANCE", + "type": "string", + "value": "0.001" + }, + { + "description": "关节数据 max(range_per_dim) 小于该阈值时判定卡死", + "key": "JOINT_TOLERANCE", + "type": "string", + "value": "0.001" + }, + { + "description": "是否要求存在 puppet/joint_position、puppet/joint_position_left 或 franka/joint_position", + "key": "REQUIRE_JOINT_POSITION", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求存在 puppet/end_effector、puppet/end_effector_left 或 franka/end_effector;默认仅在存在时检查质量", + "key": "REQUIRE_END_EFFECTOR", + "type": "boolean", + "value": "false" + }, + { + "description": "是否要求 HDF5 中存在 rgb/image/camera 相关观测数据集", + "key": "REQUIRE_IMAGES", + "type": "boolean", + "value": "false" + } + ], + "id": "operator-builtin-robomind-hdf5-quality-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-robomind-hdf5-quality-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-robomind-hdf5-quality-validate@sha256:e0973f8631bf7234ab5a0073f50f3fb79fc1cf333d0e8060596c218168f2913b", + "inputs": [ + { + "description": "RoboMIND trajectory.hdf5 文件或包含 h5_* 轨迹目录的根目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "RoboMIND HDF5质量校验", + "outputs": [ + { + "description": "RoboMIND HDF5 轨迹质量校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_robomind_hdf5_quality_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-robomind-hdf5-quality-validate", + "operatorSlug": "robomind_hdf5_quality_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "robomind-trajectory-hdf5-files", + "h5py", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_robomind_hdf5_quality_validate", + "runtimeSource": "backend/operators/builtin/robomind_hdf5_quality_validate/operator.py", + "skillDefinition": "面向 RoboMIND、Multi-embodiment Intelligence、真实遥操作 VLA 训练数据,读取 h5_franka_1rgb/h5_franka_3rgb/h5_ur_1rgb/h5_agilex_3rgb/h5_tienkung 等目录下的 trajectory.hdf5,解析 robot/task/split/episode 元信息,检查 puppet/end_effector、puppet/joint_position、master/joint_position、franka/* 候选字段的 shape、finite 数值、unique ratio 和卡死状态,并统计 RGB/image/camera 与 action/gripper/command 数据集。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 robomimic/DROID 风格 HDF5 示教文件,输出 demo 数、样本数、动作形状、观测键和 mask 信息。", + "hyperparams": [ + { + "description": ".hdf5/.h5 文件或包含 HDF5 文件的目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多返回多少个 demo 的摘要,必须为正整数", + "key": "MAX_DEMOS", + "type": "string", + "value": "50" + } + ], + "id": "operator-builtin-hdf5-demonstration-summary", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-hdf5-demonstration-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-hdf5-demonstration-summary@sha256:3d82e565a006c7c300d45ebfffaa1098d1d6b3e8bcdffd1b8a361d0768b7e08f", + "inputs": [ + { + "description": "HDF5 文件或目录输入", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "HDF5 示教数据概览", + "outputs": [ + { + "description": "HDF5 示教数据概览", + "id": "out-1", + "name": "summary", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_hdf5_demonstration_summary", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-hdf5-demonstration-summary", + "operatorSlug": "hdf5_demonstration_summary", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "hdf5-files", + "h5py" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_hdf5_demonstration_summary", + "runtimeSource": "backend/operators/builtin/hdf5_demonstration_summary/operator.py", + "skillDefinition": "面向 robomimic、DROID 等常见 HDF5 示教数据,真实读取 /data/demo_*、actions、states、obs、next_obs、mask 和 env_args 元数据;缺少 h5py 时直接失败。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 robomimic HDF5 示教文件,校验 /data/demo_* 下 actions、rewards、dones、obs、next_obs、states、mask 与 total 样本数一致性。", + "hyperparams": [ + { + "description": ".hdf5/.h5 文件或包含 robomimic HDF5 文件的目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "是否要求每个 demo 包含 rewards 数据集", + "key": "REQUIRE_REWARDS", + "type": "string", + "value": "true" + }, + { + "description": "是否要求每个 demo 包含 dones 数据集", + "key": "REQUIRE_DONES", + "type": "string", + "value": "true" + }, + { + "description": "是否要求每个 demo 包含 next_obs group", + "key": "REQUIRE_NEXT_OBS", + "type": "string", + "value": "true" + }, + { + "description": "是否检查 action 绝对值不超过 MAX_ABS_ACTION;默认关闭以兼容未归一化控制量", + "key": "CHECK_ACTION_RANGE", + "type": "string", + "value": "false" + }, + { + "description": "开启 CHECK_ACTION_RANGE 后允许的最大 action 绝对值", + "key": "MAX_ABS_ACTION", + "type": "string", + "value": "1.0" + }, + { + "description": "每个文件最多校验多少个 demo,必须为正整数", + "key": "MAX_DEMOS", + "type": "string", + "value": "200" + }, + { + "description": "最多校验多少个 HDF5 文件,必须为正整数", + "key": "MAX_FILES", + "type": "string", + "value": "20" + } + ], + "id": "operator-builtin-robomimic-hdf5-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-robomimic-hdf5-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-robomimic-hdf5-validate@sha256:0d6c7b88a5c3139f3087deb754dbc7642dcbcd2899c1c708a21b1f7c21a04e66", + "inputs": [ + { + "description": "robomimic .hdf5/.h5 文件或目录输入", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "robomimic HDF5 示教校验", + "outputs": [ + { + "description": "robomimic HDF5 示教结构校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_robomimic_hdf5_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-robomimic-hdf5-validate", + "operatorSlug": "robomimic_hdf5_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "robomimic-hdf5-files", + "h5py", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_robomimic_hdf5_validate", + "runtimeSource": "backend/operators/builtin/robomimic_hdf5_validate/operator.py", + "skillDefinition": "面向 robomimic、D4RL/robosuite 派生示教数据和 GitHub 常见机器人模仿学习 HDF5 数据,真实读取 /data/demo_* group,校验 actions/rewards/dones 与 obs/next_obs 帧数、num_samples、data.attrs.total 和 mask/train/valid 引用关系。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 ACT/ALOHA 风格 HDF5 示教文件,校验 action、observations/qpos、observations/qvel 和多相机图像帧数一致性。", + "hyperparams": [ + { + "description": ".hdf5/.h5 文件或包含 ALOHA HDF5 文件的目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "HDF5 中动作数组路径", + "key": "ACTION_PATH", + "type": "string", + "value": "/action" + }, + { + "description": "HDF5 中关节位置数组路径", + "key": "QPOS_PATH", + "type": "string", + "value": "/observations/qpos" + }, + { + "description": "HDF5 中关节速度数组路径", + "key": "QVEL_PATH", + "type": "string", + "value": "/observations/qvel" + }, + { + "description": "HDF5 中多相机图像 group 路径", + "key": "IMAGE_GROUP_PATH", + "type": "string", + "value": "/observations/images" + }, + { + "description": "是否要求 qvel 数据集必须存在", + "key": "REQUIRE_QVEL", + "type": "string", + "value": "false" + }, + { + "description": "最多校验多少个 HDF5 文件", + "key": "MAX_FILES", + "type": "string", + "value": "20" + } + ], + "id": "operator-builtin-aloha-hdf5-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-aloha-hdf5-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-aloha-hdf5-validate@sha256:e90db43b9bf900fec24f36546724121bab2a329f9ab7b48f579e84a53eb045a8", + "inputs": [ + { + "description": "ALOHA/ACT .hdf5/.h5 文件或目录输入", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "ALOHA HDF5 示教校验", + "outputs": [ + { + "description": "ALOHA HDF5 示教结构校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_aloha_hdf5_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-aloha-hdf5-validate", + "operatorSlug": "aloha_hdf5_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "aloha-act-hdf5-files", + "h5py", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_aloha_hdf5_validate", + "runtimeSource": "backend/operators/builtin/aloha_hdf5_validate/operator.py", + "skillDefinition": "面向 ALOHA、Mobile ALOHA、ACT 等双臂示教数据,真实读取 /action、/observations/qpos、/observations/qvel 和 /observations/images/ HDF5 数据集,校验缺失数据、帧数漂移、相机数量和 action/qpos 维度。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 UMI dataset.zarr 或 dataset.zarr.zip,校验 data 数组、meta/episode_ends、首维时间步一致性、关键末端/夹爪/相机字段和相机 chunk 配置。", + "hyperparams": [ + { + "description": "UMI dataset.zarr 目录、dataset.zarr.zip 文件或包含它们的目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多校验并返回多少个 data array 摘要,必须为正整数", + "key": "MAX_ARRAYS", + "type": "string", + "value": "100" + }, + { + "description": "非图像数值数组最多抽样多少帧用于 finite/min/max 检查", + "key": "MAX_SAMPLE_ROWS", + "type": "string", + "value": "256" + }, + { + "description": "逗号分隔的 UMI data 关键数组名;为空时使用默认末端位姿和夹爪宽度字段", + "key": "REQUIRED_DATA_KEYS", + "type": "string", + "value": "robot0_eef_pos,robot0_eef_rot_axis_angle,robot0_gripper_width" + }, + { + "description": "是否要求存在 camera*_rgb 形态的 RGB 图像数组", + "key": "REQUIRE_CAMERA_RGB", + "type": "boolean", + "value": "true" + }, + { + "description": "是否检查相机数组时间维 chunk 不超过 MAX_CAMERA_CHUNK_FRAMES", + "key": "CHECK_CHUNKING", + "type": "boolean", + "value": "true" + }, + { + "description": "相机 RGB 数组允许的最大时间维 chunk 帧数;UMI 官方示例为 1", + "key": "MAX_CAMERA_CHUNK_FRAMES", + "type": "string", + "value": "4" + }, + { + "description": "产物中最多返回多少个 episode_ends 值;校验仍读取完整数组", + "key": "MAX_EPISODE_ENDS", + "type": "string", + "value": "100" + } + ], + "id": "operator-builtin-umi-zarr-dataset-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-umi-zarr-dataset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-umi-zarr-dataset-validate@sha256:a6f4a48ad0f019a3720905ccbc5b8ab733baf6cf03cb0163695205d792831ff1", + "inputs": [ + { + "description": "UMI dataset.zarr 目录、dataset.zarr.zip 文件或其上级目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "UMI Zarr数据集校验", + "outputs": [ + { + "description": "UMI Zarr 数据集结构和关键训练数组校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_umi_zarr_dataset_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-umi-zarr-dataset-validate", + "operatorSlug": "umi_zarr_dataset_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "umi-zarr-dataset", + "zarr", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_umi_zarr_dataset_validate", + "runtimeSource": "backend/operators/builtin/umi_zarr_dataset_validate/operator.py", + "skillDefinition": "面向 UMI、Universal Manipulation Interface、Diffusion Policy Zarr 训练数据和社区 UMI-like 数据,真实打开 dataset.zarr 或 dataset.zarr.zip,读取 root/data 下 camera*_rgb、robot0_eef_pos、robot0_eef_rot_axis_angle、robot0_gripper_width 等数组,以及 root/meta/episode_ends,校验 episode 边界、所有 data 数组首维时间步一致、关键字段存在、数值 finite 和相机时间维 chunk 是否适合随机读取训练。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 Zarr group/array store,输出 actions、states、episode_ends 等数组形状、chunk 和 dtype 概览。", + "hyperparams": [ + { + "description": "Zarr group/array store 根目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多返回多少个 array 摘要,必须为正整数", + "key": "MAX_ARRAYS", + "type": "string", + "value": "100" + } + ], + "id": "operator-builtin-zarr-trajectory-summary", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-zarr-trajectory-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-zarr-trajectory-summary@sha256:10984c9423a77f6391516f96ad9dd6c708a1d7b0db80a87d5277165513fee505", + "inputs": [ + { + "description": "Zarr 目录或上游目录 artifact", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "Zarr 轨迹数组概览", + "outputs": [ + { + "description": "Zarr 轨迹数组概览", + "id": "out-1", + "name": "summary", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_zarr_trajectory_summary", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-zarr-trajectory-summary", + "operatorSlug": "zarr_trajectory_summary", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "zarr-v2-v3-metadata", + "filesystem" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_zarr_trajectory_summary", + "runtimeSource": "backend/operators/builtin/zarr_trajectory_summary/operator.py", + "skillDefinition": "面向 Diffusion Policy、ManiSkill、内部机器人数组轨迹等 Zarr 数据,真实读取 .zgroup/.zarray/.zattrs 或 zarr.json 元数据,汇总数组 shape、chunk、dtype、episode_count 和 step_count。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 RLBench task/variation/episode 目录,校验 low_dim_obs.pkl、variation_descriptions.pkl 和多相机 RGB/depth/mask 帧数量一致性。", + "hyperparams": [ + { + "description": "RLBench episode/task/variation 目录或其中任意文件路径", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "需要扫描的相机目录名,逗号分隔", + "key": "CAMERA_DIRS", + "type": "string", + "value": "front_rgb,left_shoulder_rgb,right_shoulder_rgb,wrist_rgb,overhead_rgb,front_depth,left_shoulder_depth,right_shoulder_depth,wrist_depth,front_mask,left_shoulder_mask,right_shoulder_mask,wrist_mask" + }, + { + "description": "最多校验多少个 episode 目录,必须为正整数", + "key": "MAX_EPISODES", + "type": "string", + "value": "200" + } + ], + "id": "operator-builtin-rlbench-dataset-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rlbench-dataset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rlbench-dataset-validate@sha256:f33e00d9a03bad99b9abeac51c69bf561b766b64188c744a154086364bc75764", + "inputs": [ + { + "description": "RLBench task、variation 或 episode 目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "RLBench 数据集校验", + "outputs": [ + { + "description": "RLBench 数据集校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_rlbench_dataset_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-rlbench-dataset-validate", + "operatorSlug": "rlbench_dataset_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "rlbench-episode-files", + "pickle-files", + "image-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_rlbench_dataset_validate", + "runtimeSource": "backend/operators/builtin/rlbench_dataset_validate/operator.py", + "skillDefinition": "面向 RLBench、RVT、PerAct 等常见多任务仿真演示数据,真实读取 episode 下 low_dim_obs.pkl 和 variation_descriptions.pkl,并扫描 front/left_shoulder/right_shoulder/wrist 等相机 RGB、depth、mask 目录,检查低维观测步数、语言描述和图像帧数是否对齐。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 ManiSkill .h5/.hdf5 轨迹和同名 JSON metadata,校验 trajectory group、actions、rewards/terminated/env_states 长度和 metadata elapsed_steps。", + "hyperparams": [ + { + "description": "ManiSkill .h5/.hdf5 轨迹文件或包含轨迹文件的目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多读取多少个 HDF5 文件", + "key": "MAX_FILES", + "type": "string", + "value": "50" + }, + { + "description": "每个 actions dataset 最多扫描多少行用于有限值检查", + "key": "MAX_ACTION_ROWS", + "type": "string", + "value": "200000" + } + ], + "id": "operator-builtin-maniskill-trajectory-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-maniskill-trajectory-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-maniskill-trajectory-validate@sha256:d18459e85b43df78ed7df591c2ecfe86122ecb927f429cf65b465c5096edf453", + "inputs": [ + { + "description": "ManiSkill HDF5 轨迹文件或目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "ManiSkill 轨迹校验", + "outputs": [ + { + "description": "ManiSkill 轨迹校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_maniskill_trajectory_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-maniskill-trajectory-validate", + "operatorSlug": "maniskill_trajectory_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "maniskill-hdf5-json-files", + "h5py", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_maniskill_trajectory_validate", + "runtimeSource": "backend/operators/builtin/maniskill_trajectory_validate/operator.py", + "skillDefinition": "面向 ManiSkill2/3、SAPIEN 仿真演示和 GitHub 常见 HDF5 轨迹数据,真实读取 .h5/.hdf5 里的 traj_* group、actions、rewards、terminated、env_states 等 dataset,并读取同名 JSON metadata 的 env_info/episodes,检查动作步数、轨迹数量和 metadata elapsed_steps 是否一致。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 CALVIN episode_*.npz 和 lang_annotations/auto_lang_ann.npy,校验 actions、rel_actions、robot_obs、scene_obs、rgb_static、rgb_gripper 等数组。", + "hyperparams": [ + { + "description": "CALVIN 数据集目录或 episode .npz 文件", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "必需 NPZ 数组名称,逗号分隔", + "key": "REQUIRED_ARRAYS", + "type": "string", + "value": "actions,rel_actions,robot_obs,scene_obs,rgb_static,rgb_gripper" + }, + { + "description": "最多读取多少个 episode .npz 文件", + "key": "MAX_FILES", + "type": "string", + "value": "500" + } + ], + "id": "operator-builtin-calvin-episode-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-calvin-episode-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-calvin-episode-validate@sha256:bc90a289bb258dc0f74fd1a33d2baa5ca53182e78eed23c42106205325ae9a48", + "inputs": [ + { + "description": "CALVIN 数据集目录或 episode .npz 文件", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "CALVIN Episode 校验", + "outputs": [ + { + "description": "CALVIN Episode 校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_calvin_episode_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-calvin-episode-validate", + "operatorSlug": "calvin_episode_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "calvin-episode-npz-files", + "numpy", + "language-annotation-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_calvin_episode_validate", + "runtimeSource": "backend/operators/builtin/calvin_episode_validate/operator.py", + "skillDefinition": "面向 CALVIN、语言条件长程操作和多相机桌面机器人数据,真实读取 episode_*.npz 中 actions、rel_actions、robot_obs、scene_obs、rgb_static、rgb_gripper 等数组,检查必需数组存在、shape/dtype、有限数值和图像数组维度,并解析 lang_annotations/auto_lang_ann.npy 的语言标注数量。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 CSV/JSON/JSONL/Parquet/HDF5 动作数据,按 RT-X/Open X 常见 7 维动作空间校验 action 维度、gripper 范围、有限值和动作幅度。", + "hyperparams": [ + { + "description": "真实 CSV/JSON/JSONL/Parquet/HDF5 动作文件或目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "候选动作字段,支持点路径,逗号分隔", + "key": "ACTION_FIELDS", + "type": "string", + "value": "action,actions,world_vector,rotation_delta,gripper_closedness_action" + }, + { + "description": "期望动作向量维度,RT-X 常见为 7", + "key": "EXPECTED_ACTION_DIM", + "type": "number", + "value": "7" + }, + { + "description": "gripper 维度索引,默认第 7 维", + "key": "GRIPPER_INDEX", + "type": "number", + "value": "6" + }, + { + "description": "gripper 允许最小值", + "key": "GRIPPER_MIN", + "type": "number", + "value": "-1" + }, + { + "description": "gripper 允许最大值", + "key": "GRIPPER_MAX", + "type": "number", + "value": "1" + }, + { + "description": "动作每个维度允许的最大绝对值", + "key": "MAX_ABS_DELTA", + "type": "number", + "value": "10" + }, + { + "description": "是否要求 gripper 维度存在且在范围内", + "key": "REQUIRE_GRIPPER", + "type": "boolean", + "value": "true" + }, + { + "description": "最多扫描的数据文件数", + "key": "MAX_FILES", + "type": "number", + "value": "50" + }, + { + "description": "最多扫描的动作行数", + "key": "MAX_ROWS", + "type": "number", + "value": "200000" + } + ], + "id": "operator-builtin-rtx-action-schema-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rtx-action-schema-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rtx-action-schema-validate@sha256:65dc02ffb329e3df1a0cadf66dc92379c92b3e98b6d35bfe460f6ea893aea031", + "inputs": [ + { + "description": "可选上游数据;通常使用 SOURCE_PATH 指向真实动作文件或目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "RT-X动作空间Schema校验", + "outputs": [ + { + "description": "RT-X 动作空间校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_rtx_action_schema_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-rtx-action-schema-validate", + "operatorSlug": "rtx_action_schema_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "csv-json-jsonl-parquet-hdf5-action-files", + "pyarrow", + "h5py", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_rtx_action_schema_validate", + "runtimeSource": "backend/operators/builtin/rtx_action_schema_validate/operator.py", + "skillDefinition": "面向 Open X-Embodiment / RT-X / BridgeData 等统一动作空间训练前验收,检查动作向量是否符合配置维度,默认 7 维末位 gripper,并约束 gripper 与 delta 幅度。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 CSV/JSON/JSONL/Parquet 元信息或 manifest,校验 robot/embodiment、task、camera、action_dim 等字段并统计多机器人覆盖。", + "hyperparams": [ + { + "description": "真实元信息或 manifest 文件/目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "必须存在的语义字段,可包含 robot_type、task、camera、action_dim", + "key": "REQUIRED_FIELDS", + "type": "string", + "value": "robot_type,task,action_dim" + }, + { + "description": "期望至少覆盖多少种机器人类型", + "key": "MIN_ROBOT_TYPES", + "type": "number", + "value": "1" + }, + { + "description": "候选机器人/embodiment 字段,逗号分隔", + "key": "ROBOT_FIELDS", + "type": "string", + "value": "robot_type,robotType,robot,embodiment,embodiment_name,platform" + }, + { + "description": "候选任务/语言指令字段,逗号分隔", + "key": "TASK_FIELDS", + "type": "string", + "value": "task,task_name,taskName,language_instruction,instruction" + }, + { + "description": "候选相机字段,逗号分隔", + "key": "CAMERA_FIELDS", + "type": "string", + "value": "camera_names,cameras,camera,observation.images,image_modalities" + }, + { + "description": "候选动作维度字段,逗号分隔", + "key": "ACTION_DIM_FIELDS", + "type": "string", + "value": "action_dim,actionDimension,action_space_dim,action_shape" + }, + { + "description": "最多扫描的元信息文件数", + "key": "MAX_FILES", + "type": "number", + "value": "50" + }, + { + "description": "最多扫描的元信息行数", + "key": "MAX_ROWS", + "type": "number", + "value": "100000" + } + ], + "id": "operator-builtin-embodiment-metadata-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-embodiment-metadata-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-embodiment-metadata-validate@sha256:47df5e78b0fd62bf046f8340b7af5d4cef3bc38efaa2b10a3a7ee028adc62ebc", + "inputs": [ + { + "description": "可选上游数据;通常使用 SOURCE_PATH 指向真实元信息或 manifest", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "多机器人Embodiment元信息校验", + "outputs": [ + { + "description": "Embodiment 元信息校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_embodiment_metadata_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-embodiment-metadata-validate", + "operatorSlug": "embodiment_metadata_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "csv-json-jsonl-parquet-metadata-files", + "pyarrow" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_embodiment_metadata_validate", + "runtimeSource": "backend/operators/builtin/embodiment_metadata_validate/operator.py", + "skillDefinition": "面向 RoboMIND、Open X-Embodiment 等多机器人数据集,统一验收机器人类型、任务、相机命名和动作维度元信息,避免混合数据训练前 metadata 不可追溯。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 CSV/JSON/JSONL/Parquet/HDF5 力/扭矩/wrench 数据,校验 3D/6D 维度、有限值、阈值范围和时间戳单调性。", + "hyperparams": [ + { + "description": "真实 CSV/JSON/JSONL/Parquet/HDF5 力/扭矩传感器文件或目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "期望力/扭矩向量维度,常见 wrench 为 6", + "key": "EXPECTED_DIM", + "type": "number", + "value": "6" + }, + { + "description": "前三个 force 维度的最大绝对值阈值", + "key": "MAX_FORCE_ABS", + "type": "number", + "value": "10000" + }, + { + "description": "torque 维度的最大绝对值阈值", + "key": "MAX_TORQUE_ABS", + "type": "number", + "value": "10000" + }, + { + "description": "是否校验时间戳单调不回退", + "key": "CHECK_TIMESTAMP_MONOTONIC", + "type": "boolean", + "value": "true" + }, + { + "description": "候选 wrench 字段,逗号分隔", + "key": "WRENCH_FIELDS", + "type": "string", + "value": "wrench,force_torque,forceTorque,ft,sensor.wrench" + }, + { + "description": "候选 force 字段,逗号分隔", + "key": "FORCE_FIELDS", + "type": "string", + "value": "force,forces,sensor.force" + }, + { + "description": "候选 torque 字段,逗号分隔", + "key": "TORQUE_FIELDS", + "type": "string", + "value": "torque,torques,sensor.torque" + }, + { + "description": "候选时间戳字段,逗号分隔", + "key": "TIMESTAMP_FIELDS", + "type": "string", + "value": "timestamp,time,t,frame_timestamp,sensor.timestamp" + }, + { + "description": "最多扫描的数据文件数", + "key": "MAX_FILES", + "type": "number", + "value": "50" + }, + { + "description": "最多扫描的数据行数", + "key": "MAX_ROWS", + "type": "number", + "value": "200000" + } + ], + "id": "operator-builtin-force-torque-sensor-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-force-torque-sensor-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-force-torque-sensor-validate@sha256:d7e4558fd8fad892f328af10f21eaa8ca3f8122fd7fe209184b99c5f51a7a22a", + "inputs": [ + { + "description": "可选上游数据;通常使用 SOURCE_PATH 指向真实力/扭矩传感器文件或目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "力扭矩传感器数据校验", + "outputs": [ + { + "description": "力/扭矩传感器校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_force_torque_sensor_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-force-torque-sensor-validate", + "operatorSlug": "force_torque_sensor_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "csv-json-jsonl-parquet-hdf5-force-torque-files", + "pyarrow", + "h5py", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_force_torque_sensor_validate", + "runtimeSource": "backend/operators/builtin/force_torque_sensor_validate/operator.py", + "skillDefinition": "面向接触丰富操作、失败检测、装配和遥操作回放数据,真实读取 wrench/force/torque 序列,检查传感器数据是否可训练、可同步、未饱和。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 episode/step JSON 或 JSONL,校验动作、状态、图像引用等多模态字段完整性和维度一致性。", + "hyperparams": [ + { + "description": "episode JSON/JSONL 文件或包含 episodes.jsonl 的目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "逗号分隔的必需字段路径,支持 observation.image 等点路径", + "key": "REQUIRED_FIELDS", + "type": "string", + "value": "action,observation.image,observation.state" + }, + { + "description": "是否校验 observation.image 等相对路径在数据目录中真实存在", + "key": "CHECK_FILE_EXISTS", + "type": "string", + "value": "true" + }, + { + "description": "是否要求所有 step 的 action 维度一致", + "key": "REQUIRE_CONSISTENT_ACTION_DIM", + "type": "string", + "value": "true" + }, + { + "description": "是否要求所有 step 的 observation.state 维度一致", + "key": "REQUIRE_CONSISTENT_STATE_DIM", + "type": "string", + "value": "true" + } + ], + "id": "operator-builtin-trajectory-integrity-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-trajectory-integrity-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-trajectory-integrity-validate@sha256:1ec455b21a6a4cf3c35953d50d8965c77deeffd71412ffa1ebb3bc66f42fbe9e", + "inputs": [ + { + "description": "包含 episodes/steps 的轨迹 JSON/JSONL 或上游 artifact", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "轨迹多模态完整性校验", + "outputs": [ + { + "description": "轨迹多模态完整性校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_trajectory_integrity_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-trajectory-integrity-validate", + "operatorSlug": "trajectory_integrity_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "trajectory-json", + "local-media-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_trajectory_integrity_validate", + "runtimeSource": "backend/operators/builtin/trajectory_integrity_validate/operator.py", + "skillDefinition": "面向 Open X-Embodiment、BridgeData、DROID、LeRobot 等具身轨迹数据,逐 step 校验 action、observation、image/state 等字段和本地文件引用,输出可审计缺失项。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 JSON/JSONL/Parquet 轨迹表,校验 action/state 缺失、维度一致、非有限值和异常幅值。", + "hyperparams": [ + { + "description": "轨迹 JSON/JSONL/Parquet 文件或目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "候选动作字段,逗号分隔;支持 observation.state 这样的点路径", + "key": "ACTION_FIELDS", + "type": "string", + "value": "action,actions,control,command" + }, + { + "description": "候选状态/本体感知字段,逗号分隔", + "key": "STATE_FIELDS", + "type": "string", + "value": "observation.state,state,states,observation.joint_positions,joint_positions,proprio" + }, + { + "description": "最多读取多少行真实数据", + "key": "MAX_ROWS", + "type": "string", + "value": "10000" + }, + { + "description": "最多读取多少个输入文件", + "key": "MAX_FILES", + "type": "string", + "value": "50" + }, + { + "description": "action/state 元素允许的最大绝对值,超过会记为异常幅值", + "key": "MAX_ABS_VALUE", + "type": "string", + "value": "1000000" + }, + { + "description": "维度低方差提示阈值;仅作为 warning,不直接判失败", + "key": "LOW_VARIANCE_EPSILON", + "type": "string", + "value": "0.000000001" + } + ], + "id": "operator-builtin-action-state-quality-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-action-state-quality-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-action-state-quality-validate@sha256:96055934c67458386848e774725ea4328cb0b1fe157e77a23e8586659e7b538d", + "inputs": [ + { + "description": "包含 action/state 的轨迹 JSON、JSONL、Parquet 文件或目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "动作状态数值质量校验", + "outputs": [ + { + "description": "动作状态数值质量校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_action_state_quality_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-action-state-quality-validate", + "operatorSlug": "action_state_quality_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-jsonl-parquet-files", + "pyarrow" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_action_state_quality_validate", + "runtimeSource": "backend/operators/builtin/action_state_quality_validate/operator.py", + "skillDefinition": "面向 LeRobot、DROID、Open X-Embodiment/RLDS 导出的训练表,真实读取 action、observation.state、joint_positions 等字段,检查缺失、非数字、NaN/Inf、维度漂移和异常大数值,并输出每个维度的 min/max/mean。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 episode JSON/JSONL/Parquet,按 episode 分组校验 frame_index 连续性和终止标志位置。", + "hyperparams": [ + { + "description": "episode JSON/JSONL/Parquet 文件或目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "候选 episode ID 字段,逗号分隔", + "key": "EPISODE_FIELDS", + "type": "string", + "value": "episode_index,episode_id,episodeId,trajectory_id,trajectoryId" + }, + { + "description": "候选帧/步索引字段,逗号分隔", + "key": "FRAME_FIELDS", + "type": "string", + "value": "frame_index,step_index,timestep,time_step" + }, + { + "description": "候选终止标志字段,逗号分隔", + "key": "TERMINAL_FIELDS", + "type": "string", + "value": "is_last,is_terminal,done,terminated,success" + }, + { + "description": "是否要求每个 episode 的第一帧索引为 0", + "key": "REQUIRE_ZERO_FRAME_START", + "type": "string", + "value": "true" + }, + { + "description": "是否要求每个 episode 最后一帧有终止标志", + "key": "REQUIRE_TERMINAL_FLAG", + "type": "string", + "value": "true" + }, + { + "description": "最多读取多少行真实数据", + "key": "MAX_ROWS", + "type": "string", + "value": "100000" + }, + { + "description": "最多读取多少个输入文件", + "key": "MAX_FILES", + "type": "string", + "value": "50" + } + ], + "id": "operator-builtin-episode-boundary-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-episode-boundary-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-episode-boundary-validate@sha256:9e7782f3fe626f858138fd066ca2911d02cc4bfe6ab933065b4ebfedaff544ba", + "inputs": [ + { + "description": "包含 episode/frame/done 字段的轨迹文件或目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "Episode 边界校验", + "outputs": [ + { + "description": "Episode 边界校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_episode_boundary_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-episode-boundary-validate", + "operatorSlug": "episode_boundary_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-jsonl-parquet-files", + "pyarrow" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_episode_boundary_validate", + "runtimeSource": "backend/operators/builtin/episode_boundary_validate/operator.py", + "skillDefinition": "面向 RLDS、LeRobot、DROID、ALOHA 等 episode 化轨迹,真实读取 episode_index、frame_index、is_last/done/terminated/success 等字段,检查每个 episode 是否从 0 开始、frame 是否连续递增、终止标志是否只出现在末尾。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 RGB 与 depth 图像目录,按帧名配对,校验尺寸一致性、深度有效像素、深度范围和相机内参。", + "hyperparams": [ + { + "description": "RGB-D 数据集根目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "可选 RGB 目录;为空时自动从 rgb/color/images 推断", + "key": "RGB_DIR", + "type": "string", + "value": "" + }, + { + "description": "可选 depth 目录;为空时自动从 depth/depths/images/depth 推断", + "key": "DEPTH_DIR", + "type": "string", + "value": "" + }, + { + "description": "可选 intrinsics JSON/YAML;为空时自动查找 camera_intrinsics.json/intrinsics.json/camera.yaml", + "key": "INTRINSICS_PATH", + "type": "string", + "value": "" + }, + { + "description": "depth 原始单位到米的缩放分母,例如 uint16 毫米深度使用 1000", + "key": "DEPTH_SCALE", + "type": "string", + "value": "1000" + }, + { + "description": "最多校验多少对 RGB-D 帧", + "key": "MAX_PAIRS", + "type": "string", + "value": "200" + } + ], + "id": "operator-builtin-rgbd-frame-pair-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rgbd-frame-pair-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rgbd-frame-pair-validate@sha256:4c43bb4d7b81eea7604f3dce8d074201de8187aa2b12cce1e2f7fa1f049838b8", + "inputs": [ + { + "description": "RGB-D 数据集目录或上游目录 artifact", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "RGB-D 帧对齐校验", + "outputs": [ + { + "description": "RGB-D 帧对齐校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_rgbd_frame_pair_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-rgbd-frame-pair-validate", + "operatorSlug": "rgbd_frame_pair_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "rgb-images", + "depth-images", + "opencv-python-headless", + "numpy", + "pyyaml" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_rgbd_frame_pair_validate", + "runtimeSource": "backend/operators/builtin/rgbd_frame_pair_validate/operator.py", + "skillDefinition": "面向 ManiSkill、真实机器人 RGB-D 采集、仿真深度图和 VLA 视觉观测入库,真实读取 rgb/color/images 与 depth/depths PNG/TIFF/NPY,校验同名帧是否成对、图像尺寸是否一致、depth 是否有有效像素,并解析 fx/fy/cx/cy 内参。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实语义/实例分割 mask 与可选原图,校验 mask/图像配对、尺寸一致性、类别 ID 范围和像素分布。", + "hyperparams": [ + { + "description": "分割数据集根目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "可选原图目录;为空时从 images/rgb/color 推断", + "key": "IMAGE_DIR", + "type": "string", + "value": "" + }, + { + "description": "可选 mask 目录;为空时从 masks/segmentation/labels/semantic 推断", + "key": "MASK_DIR", + "type": "string", + "value": "" + }, + { + "description": "可选允许的类别 ID,逗号分隔;为空时只统计不限制", + "key": "CLASS_IDS", + "type": "string", + "value": "" + }, + { + "description": "最多校验多少个 mask 或图像/mask 对", + "key": "MAX_PAIRS", + "type": "string", + "value": "200" + } + ], + "id": "operator-builtin-segmentation-mask-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-segmentation-mask-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-segmentation-mask-validate@sha256:c96e3c47fe7bab282925c75d473fd1e8c400605c5d33b9dac9ab68ca98e4fbd3", + "inputs": [ + { + "description": "分割 mask 数据集目录或上游目录 artifact", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "分割 Mask 校验", + "outputs": [ + { + "description": "分割 mask 校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_segmentation_mask_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-segmentation-mask-validate", + "operatorSlug": "segmentation_mask_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "segmentation-mask-images", + "opencv-python-headless", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_segmentation_mask_validate", + "runtimeSource": "backend/operators/builtin/segmentation_mask_validate/operator.py", + "skillDefinition": "面向仿真语义分割、机器人场景理解、抓取区域 mask 和实例分割训练数据,真实读取 masks/segmentation/labels/semantic PNG/TIFF/NPY,统计 class id 像素分布,校验图像/Mask 成对关系与尺寸一致性。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 JSONL/CSV/Parquet/JSON 元数据中的 language_instruction/instruction/task 等字段,统计覆盖率、去重和任务分布。", + "hyperparams": [ + { + "description": "元数据文件或包含 episodes/tasks 的目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "逗号分隔的候选语言字段名", + "key": "TEXT_FIELDS", + "type": "string", + "value": "language_instruction,instruction,task,command,prompt" + }, + { + "description": "逗号分隔的 episode/task 分组字段名", + "key": "GROUP_FIELDS", + "type": "string", + "value": "episode_index,episode_id,trajectory_id,task_index,task_id" + }, + { + "description": "最多读取多少行元数据", + "key": "MAX_ROWS", + "type": "string", + "value": "10000" + } + ], + "id": "operator-builtin-language-instruction-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-language-instruction-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-language-instruction-validate@sha256:302e6fee95108b952abdbd080e18641c08717983a0fa6b2efdcf387ba25e2249", + "inputs": [ + { + "description": "包含语言指令的 episode/task 元数据文件或目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "语言指令校验", + "outputs": [ + { + "description": "语言指令覆盖与质量校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_language_instruction_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-language-instruction-validate", + "operatorSlug": "language_instruction_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-jsonl-csv-parquet-files", + "pyarrow" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_language_instruction_validate", + "runtimeSource": "backend/operators/builtin/language_instruction_validate/operator.py", + "skillDefinition": "面向 Open X-Embodiment、RLDS、LeRobot、VLA 训练集和人工采集任务元数据,真实读取 episodes/tasks JSONL、CSV、Parquet 或 JSON,检查 language_instruction/instruction/task/command/prompt 字段覆盖率、空指令、重复指令和 episode/task 分组数量。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 VLA/OpenVLA/DROID 风格 JSON、JSONL、CSV 或 Parquet 样本清单,校验图像引用、语言指令和动作向量完整性。", + "hyperparams": [ + { + "description": "VLA 样本 JSON/JSONL/CSV/Parquet 文件或目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "逗号分隔的候选图像引用字段,支持点路径", + "key": "IMAGE_FIELDS", + "type": "string", + "value": "image,image_path,images,observation.image,observation.images,observation.exterior_image_1_left,observation.exterior_image_2_left,observation.wrist_image_left,exterior_image_1_left,exterior_image_2_left,wrist_image_left" + }, + { + "description": "逗号分隔的候选语言字段,支持 conversations 列表", + "key": "TEXT_FIELDS", + "type": "string", + "value": "language_instruction,instruction,task,prompt,goal,conversations" + }, + { + "description": "逗号分隔的候选动作字段", + "key": "ACTION_FIELDS", + "type": "string", + "value": "action,actions,control,controls,target_action" + }, + { + "description": "是否校验本地图像路径真实存在;远程 URI 只计数不下载", + "key": "CHECK_IMAGE_EXISTS", + "type": "string", + "value": "true" + }, + { + "description": "是否要求所有 action 维度一致", + "key": "REQUIRE_CONSISTENT_ACTION_DIM", + "type": "string", + "value": "true" + }, + { + "description": "最多读取多少行真实样本", + "key": "MAX_ROWS", + "type": "string", + "value": "10000" + }, + { + "description": "最多读取多少个清单文件", + "key": "MAX_FILES", + "type": "string", + "value": "50" + } + ], + "id": "operator-builtin-vla-sample-manifest-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-vla-sample-manifest-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-vla-sample-manifest-validate@sha256:bcfb95b9a40b9152f2d102c9a00b18e157281e6d7438c84cd7ec6dfb6160df69", + "inputs": [ + { + "description": "包含图像、语言指令和动作字段的 VLA 样本清单文件或目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "VLA 样本清单校验", + "outputs": [ + { + "description": "VLA 样本清单校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_vla_sample_manifest_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-vla-sample-manifest-validate", + "operatorSlug": "vla_sample_manifest_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-jsonl-csv-parquet-files", + "image-files", + "pyarrow" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_vla_sample_manifest_validate", + "runtimeSource": "backend/operators/builtin/vla_sample_manifest_validate/operator.py", + "skillDefinition": "面向 OpenVLA、DROID、Bridge/Open X-Embodiment 派生 VLA 训练清单,真实读取 image/image_path/observation.image、language_instruction/instruction/conversations 和 action/control 字段,校验本地图像引用存在、语言指令覆盖、动作维度一致和完整样本数量。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 JSON/YAML 相机标定文件,校验内参、畸变、投影/外参矩阵和图像尺寸。", + "hyperparams": [ + { + "description": "标定 JSON/YAML 文件或包含标定文件的目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多校验多少个标定文件,必须为正整数", + "key": "MAX_FILES", + "type": "string", + "value": "50" + } + ], + "id": "operator-builtin-camera-calibration-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-camera-calibration-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-camera-calibration-validate@sha256:b43c83210cdd287d8234ff4e137586b1e5d671945ef72308246e9cceb8d7cae7", + "inputs": [ + { + "description": "camera_info/calibration/intrinsics JSON 或 YAML 文件/目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "相机标定校验", + "outputs": [ + { + "description": "相机标定校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_camera_calibration_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-camera-calibration-validate", + "operatorSlug": "camera_calibration_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "camera-calibration-files", + "json-files", + "optional-yaml-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_camera_calibration_validate", + "runtimeSource": "backend/operators/builtin/camera_calibration_validate/operator.py", + "skillDefinition": "面向机器人采集、ROS CameraInfo、OpenCV 标定和多相机数据集入库,真实读取 K/D/R/P 或 camera_matrix/distortion/extrinsic 字段,校验矩阵长度、焦距、有限数值和图像尺寸。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 ROS TransformStamped 风格 JSON/JSONL/YAML 或扁平 CSV 坐标变换,校验父子 frame、平移、四元数归一化、多父节点、重复边和环。", + "hyperparams": [ + { + "description": "坐标变换文件或包含坐标变换文件的目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "四元数 norm 与 1 的允许误差", + "key": "QUATERNION_NORM_TOLERANCE", + "type": "string", + "value": "0.01" + }, + { + "description": "最多校验多少条 transform 记录,必须为正整数", + "key": "MAX_TRANSFORMS", + "type": "string", + "value": "10000" + }, + { + "description": "最多扫描多少个 transform 文件,必须为正整数", + "key": "MAX_FILES", + "type": "string", + "value": "50" + } + ], + "id": "operator-builtin-tf-transform-tree-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-tf-transform-tree-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-tf-transform-tree-validate@sha256:4777877db8114d07f770211c4d9b3375300bcbf52823d0abd5696fd2747bb4bc", + "inputs": [ + { + "description": "transforms/tf/tf_static JSON、JSONL、CSV 或 YAML 文件/目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "TF 坐标树校验", + "outputs": [ + { + "description": "TF 坐标树校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_tf_transform_tree_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-tf-transform-tree-validate", + "operatorSlug": "tf_transform_tree_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-jsonl-csv-yaml-files", + "optional-pyyaml" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_tf_transform_tree_validate", + "runtimeSource": "backend/operators/builtin/tf_transform_tree_validate/operator.py", + "skillDefinition": "面向 ROS/ROS2 机器人数据集、相机外参和多传感器标定入库,真实读取 header.frame_id、child_frame_id、translation、rotation 或 parent_frame/child_frame/qx/qy/qz/qw,校验坐标树根、叶子、重复边、多父节点和环路。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 ROS sensor_msgs/JointState 风格 JSON/JSONL/CSV/Parquet,校验 name、position、velocity、effort 数组长度、有限数值、关节名一致性和可选时间单调性。", + "hyperparams": [ + { + "description": "JointState 文件或包含 JointState 文件的目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多读取多少条 JointState 消息", + "key": "MAX_ROWS", + "type": "string", + "value": "100000" + }, + { + "description": "最多读取多少个输入文件", + "key": "MAX_FILES", + "type": "string", + "value": "50" + }, + { + "description": "是否要求所有消息的关节名称顺序一致", + "key": "REQUIRE_CONSISTENT_NAMES", + "type": "string", + "value": "true" + }, + { + "description": "是否要求 timestamp/header.stamp 单调不降", + "key": "REQUIRE_MONOTONIC_TIME", + "type": "string", + "value": "false" + } + ], + "id": "operator-builtin-ros-joint-state-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-ros-joint-state-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-ros-joint-state-validate@sha256:55199b510c1c2329facf0a12214c17fb3c4d7c256e89b6772c355a072bd878f7", + "inputs": [ + { + "description": "JointState JSON、JSONL、CSV、Parquet 文件或目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "ROS JointState 校验", + "outputs": [ + { + "description": "ROS JointState 校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_ros_joint_state_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-ros-joint-state-validate", + "operatorSlug": "ros_joint_state_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "ros-joint-state-files", + "json-jsonl-csv-parquet-files", + "pyarrow" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_ros_joint_state_validate", + "runtimeSource": "backend/operators/builtin/ros_joint_state_validate/operator.py", + "skillDefinition": "面向 ROS/ROS2 机器人采集、rosbag 导出、控制日志和训练前关节状态入库,真实读取 sensor_msgs/JointState 风格 name/position/velocity/effort 字段,检查数组长度同 name 一致、数值有限、关节名称序列是否稳定,并可选检查 timestamp 单调。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 GraspNet-1Billion 数据根目录、scene 目录或 camera 目录,校验 scenes/scene_XXXX、RGB-D、mask、MAT/XML/rect 标注、object models、grasp_label 和 collision_label。", + "hyperparams": [ + { + "description": "GraspNet 根目录、scene 目录或 camera 目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "校验 camera:any、kinect、realsense 或逗号分隔子集", + "key": "CAMERA", + "type": "string", + "value": "any" + }, + { + "description": "最多校验多少个 scene_XXXX 目录,必须为正整数", + "key": "MAX_SCENES", + "type": "string", + "value": "20" + }, + { + "description": "每个 camera 最多校验多少帧,必须为正整数", + "key": "MAX_FRAMES_PER_CAMERA", + "type": "string", + "value": "32" + }, + { + "description": "是否要求 object_id_list 中每个对象存在 nontextured.ply、textured.obj、textured.sdf", + "key": "REQUIRE_MODELS", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求每个 object id 存在 grasp_label/%03d_labels.npz 且包含 points/offsets/scores/collision", + "key": "REQUIRE_GRASP_LABELS", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求每个 scene 存在 collision_label/scene_XXXX/collision_labels.npz", + "key": "REQUIRE_COLLISION_LABELS", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求每帧存在 rect/%04d.npy 矩形抓取标签", + "key": "REQUIRE_RECT_LABELS", + "type": "boolean", + "value": "false" + }, + { + "description": "是否要求 scene 目录存在 object_id_list.txt", + "key": "REQUIRE_OBJECT_IDS", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求 camera 目录存在 camK.npy、camera_poses.npy、cam0_wrt_table.npy", + "key": "REQUIRE_CAMERA_POSES", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求每帧存在 meta/%04d.mat", + "key": "REQUIRE_META_MAT", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求每帧存在 annotations/%04d.xml 且可解析", + "key": "REQUIRE_ANNOTATION_XML", + "type": "boolean", + "value": "true" + } + ], + "id": "operator-builtin-graspnet-dataset-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-graspnet-dataset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-graspnet-dataset-validate@sha256:c4ec92cc4e87e5df975377608f02ba54de0570d0235dd36840d4a311875c060a", + "inputs": [ + { + "description": "GraspNet 根目录、scenes/scene_XXXX 目录或 kinect/realsense camera 目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "GraspNet 抓取数据集校验", + "outputs": [ + { + "description": "GraspNet RGB-D 抓取数据集结构与标注校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_graspnet_dataset_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-graspnet-dataset-validate", + "operatorSlug": "graspnet_dataset_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "graspnet-1billion-dataset-files", + "numpy", + "npz-npy-files", + "rgb-depth-mask-png-files", + "xml-annotations" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_graspnet_dataset_validate", + "runtimeSource": "backend/operators/builtin/graspnet_dataset_validate/operator.py", + "skillDefinition": "面向 GraspNet-1Billion、AnyGrasp、SuctionNet 和通用物体抓取训练数据,真实读取 models/%03d/nontextured.ply|textured.obj|textured.sdf、grasp_label/%03d_labels.npz、collision_label/scene_XXXX/collision_labels.npz、scenes/scene_XXXX/object_id_list.txt、rs_wrt_kn.npy、kinect|realsense/camK.npy、camera_poses.npy、cam0_wrt_table.npy、rgb/depth/label PNG、meta MAT、annotations XML 和 rect NPY,校验 6DoF grasp label、碰撞标签、相机内外参、RGB-D 帧和实例 mask 是否可用于真实抓取模型训练。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 BOP scenewise 数据根目录、split/scene 目录或 scene_*.json,校验 camera、models_info、scene_camera、scene_gt、scene_gt_info、rgb/depth/mask/mask_visib 与 6D object pose 核心字段。", + "hyperparams": [ + { + "description": "BOP 数据根目录、split/scene 目录或 scene_*.json 文件", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多校验多少个 scene 目录,必须为正整数", + "key": "MAX_SCENES", + "type": "string", + "value": "20" + }, + { + "description": "每个 scene 最多校验多少个 image id,必须为正整数", + "key": "MAX_IMAGES_PER_SCENE", + "type": "string", + "value": "200" + }, + { + "description": "是否要求每个 image id 存在 rgb 或 gray 图像文件", + "key": "CHECK_IMAGE_FILES", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求每个 image id 存在 depth 图像文件;RGB-only/HOT3D 等数据可关闭", + "key": "CHECK_DEPTH_FILES", + "type": "boolean", + "value": "false" + }, + { + "description": "是否要求每个 scene_gt annotation 存在 mask 与 mask_visib 文件", + "key": "CHECK_MASK_FILES", + "type": "boolean", + "value": "false" + }, + { + "description": "是否要求 scene_gt*.json 存在并包含 object pose 注解", + "key": "REQUIRE_SCENE_GT", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求 scene_gt_info*.json 存在并与 scene_gt annotation 数量一致", + "key": "REQUIRE_SCENE_GT_INFO", + "type": "boolean", + "value": "false" + }, + { + "description": "是否要求数据根目录存在 models*/models_info.json", + "key": "REQUIRE_MODELS_INFO", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求 models_info.json 中每个 obj_id 存在对应 obj_XXXXXX.ply", + "key": "REQUIRE_MODEL_FILES", + "type": "boolean", + "value": "false" + }, + { + "description": "是否要求每个 annotation 存在 cam_R_m2c 与 cam_t_m2c 且 shape 合法", + "key": "REQUIRE_OBJECT_POSE", + "type": "boolean", + "value": "true" + } + ], + "id": "operator-builtin-bop-dataset-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-bop-dataset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-bop-dataset-validate@sha256:6b99f50e0777e02df933fb196666f6bace7c251fcdfedf1630844525905d5c1b", + "inputs": [ + { + "description": "BOP 数据根目录、split/scene 目录或单个 scene_*.json 文件", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "BOP 6D姿态数据集校验", + "outputs": [ + { + "description": "BOP 6D object pose 数据集结构与标注校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_bop_dataset_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-bop-dataset-validate", + "operatorSlug": "bop_dataset_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "bop-scenewise-dataset-files", + "json-files", + "rgb-depth-mask-image-files", + "ply-model-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_bop_dataset_validate", + "runtimeSource": "backend/operators/builtin/bop_dataset_validate/operator.py", + "skillDefinition": "面向 BOP benchmark、YCB-V、T-LESS、HOT3D、XYZ-IBD、HOPE 等 6D object pose / RGB-D 机器人感知数据,真实读取 camera.json、models*/models_info.json、train|val|test*//scene_camera*.json、scene_gt*.json、scene_gt_info*.json、rgb|gray、depth、mask、mask_visib,校验 cam_K、depth_scale、obj_id、cam_R_m2c、cam_t_m2c、bbox、visib_fract 等字段并统计 split、scene、image、object 和 annotation 覆盖。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 DexYCB 数据根目录或 subject 子目录,校验 labels_*.npz、color_*.jpg、aligned_depth_to_color_*.png、calibration/models 和手-物体姿态核心数组。", + "hyperparams": [ + { + "description": "DexYCB 数据根目录、subject 子目录、camera 目录或单个 labels_*.npz 文件", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多校验多少个 labels_*.npz 样本,必须为正整数", + "key": "MAX_SAMPLES", + "type": "string", + "value": "200" + }, + { + "description": "是否要求每个 labels_*.npz 对应 color_.jpg 与 aligned_depth_to_color_.png", + "key": "CHECK_COLOR_DEPTH_FILES", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求数据根目录包含 calibration/ 与 models/;只校验 subject 分包时可关闭", + "key": "REQUIRE_ROOT_ASSETS", + "type": "boolean", + "value": "true" + }, + { + "description": "是否检查 label 数组不含 NaN/Inf", + "key": "REQUIRE_FINITE_NUMERIC", + "type": "boolean", + "value": "true" + } + ], + "id": "operator-builtin-dexycb-dataset-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dexycb-dataset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dexycb-dataset-validate@sha256:b412bc8d84fa31ff1233f13c0758156013ec03e9cd1092e9871b8bb8b4b9e6f1", + "inputs": [ + { + "description": "DexYCB 数据根目录、subject 子目录或单个 labels_*.npz 文件", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "DexYCB 数据集校验", + "outputs": [ + { + "description": "DexYCB RGB-D 与标签结构校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_dexycb_dataset_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-dexycb-dataset-validate", + "operatorSlug": "dexycb_dataset_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "dexycb-dataset-files", + "numpy", + "npz-label-files", + "rgb-depth-image-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_dexycb_dataset_validate", + "runtimeSource": "backend/operators/builtin/dexycb_dataset_validate/operator.py", + "skillDefinition": "面向 NVIDIA DexYCB 真实手-物体抓取数据,读取 20200709-subject-01/20200709_141841//labels_000053.npz 这类结构,校验 color_*.jpg、aligned_depth_to_color_*.png、labels_*.npz 中 seg、pose_y、pose_m、joint_3d、joint_2d 的 shape/dtype/finite,统计 subject、camera、object pose 和可见 hand joints,并可要求数据根目录包含 calibration 与 models。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 HO-3D/H2O-3D 数据根目录、split/sequence 目录或 meta/*.pkl,校验 train/evaluation 下 rgb、depth、segr、meta pickle 和手-物体 3D 姿态核心字段。", + "hyperparams": [ + { + "description": "HO-3D/H2O-3D 数据根目录、split/sequence 目录或单个 meta/*.pkl 文件", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多校验多少个 meta/*.pkl 样本,必须为正整数", + "key": "MAX_SAMPLES", + "type": "string", + "value": "200" + }, + { + "description": "是否要求每个 meta/*.pkl 对应 rgb/.jpg|png 与 depth/.png", + "key": "CHECK_RGB_DEPTH_FILES", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求每个样本存在 segr/.png 分割图", + "key": "CHECK_SEGMENTATION_FILES", + "type": "boolean", + "value": "false" + }, + { + "description": "是否要求 objRot 与 objTrans 存在且 shape 合法", + "key": "REQUIRE_OBJECT_POSE", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求存在 MANO handPose/handTrans/handBeta 或 handJoints3D 手部注解", + "key": "REQUIRE_HAND_ANNOTATION", + "type": "boolean", + "value": "true" + }, + { + "description": "是否检查 meta 数组不含 NaN/Inf", + "key": "REQUIRE_FINITE_NUMERIC", + "type": "boolean", + "value": "true" + } + ], + "id": "operator-builtin-ho3d-dataset-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-ho3d-dataset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-ho3d-dataset-validate@sha256:9edf87bceb97b38907f6270835c2b9fcd6dd347d800efaf41a6d747eddd8af8c", + "inputs": [ + { + "description": "HO-3D/H2O-3D 数据根目录、split/sequence 目录或单个 meta/*.pkl 文件", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "HO3D/H2O3D 数据集校验", + "outputs": [ + { + "description": "HO-3D/H2O-3D 目录与 meta 注解校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_ho3d_dataset_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-ho3d-dataset-validate", + "operatorSlug": "ho3d_dataset_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "ho3d-h2o3d-dataset-files", + "pickle", + "numpy", + "rgb-depth-image-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_ho3d_dataset_validate", + "runtimeSource": "backend/operators/builtin/ho3d_dataset_validate/operator.py", + "skillDefinition": "面向 HO-3D v2/v3 与 H2O-3D 手-物体 3D 姿态数据,真实读取 train//rgb、depth、segr、meta/.pkl 这类结构,校验 camMat、objRot、objTrans、objCorners3DRest、objName、handPose、handTrans、handBeta、handJoints3D、handBoundingBox 和可选 contact map 字段,并统计 split、sequence、object 与手部注解覆盖。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 DexYCB/HO3D 风格手-物体姿态标注 JSON/JSONL/CSV/Parquet,校验 21 点手部 2D/3D 关键点、物体 6D pose、bbox、图像和 mask 引用。", + "hyperparams": [ + { + "description": "标注文件或包含标注文件的目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "期望手部关键点数量,DexYCB/HO3D 常用 21", + "key": "EXPECTED_HAND_JOINTS", + "type": "string", + "value": "21" + }, + { + "description": "是否要求 image/rgb 文件真实存在", + "key": "CHECK_IMAGE_EXISTS", + "type": "string", + "value": "true" + }, + { + "description": "是否要求 mask/segmentation 文件真实存在", + "key": "CHECK_MASK_EXISTS", + "type": "string", + "value": "false" + }, + { + "description": "最多读取多少条姿态标注", + "key": "MAX_ROWS", + "type": "string", + "value": "100000" + }, + { + "description": "最多读取多少个标注文件", + "key": "MAX_FILES", + "type": "string", + "value": "50" + } + ], + "id": "operator-builtin-hand-object-pose-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-hand-object-pose-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-hand-object-pose-validate@sha256:f91c6c745f098d7a592c3738b4e9bbb4a2be1604605da5ccd539ef90b010deed", + "inputs": [ + { + "description": "手-物体姿态标注文件或数据集目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "手-物体姿态标注校验", + "outputs": [ + { + "description": "手-物体姿态标注校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_hand_object_pose_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-hand-object-pose-validate", + "operatorSlug": "hand_object_pose_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-jsonl-csv-parquet-files", + "image-files", + "mask-files", + "pyarrow" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_hand_object_pose_validate", + "runtimeSource": "backend/operators/builtin/hand_object_pose_validate/operator.py", + "skillDefinition": "面向 DexYCB、HO3D、机器人手抓取和具身操作标注入库,真实读取 image/mask 路径、hand_joints_3d、hand_joints_2d、object_pose_matrix 或 quaternion+translation、bbox 等字段,检查文件存在、关键点 shape、物体位姿和 bbox 合法性。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 PartNet-Mobility/SAPIEN articulated object 资产目录,校验 mobility.urdf、semantics.txt、mobility_v2.json、meta/result/bounding box、textured_objs mesh 与 point_sample。", + "hyperparams": [ + { + "description": "PartNet-Mobility 对象目录、dataset/sapien_dataset 根目录或 mobility.urdf 文件", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多校验多少个对象目录,必须为正整数", + "key": "MAX_OBJECTS", + "type": "string", + "value": "50" + }, + { + "description": "每个对象最多返回多少个 mesh/joint 引用预览", + "key": "MAX_MESH_REFERENCES", + "type": "string", + "value": "200" + }, + { + "description": "是否要求对象存在 semantics.txt 且 link 语义可解析", + "key": "REQUIRE_SEMANTICS", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求对象存在 mobility_v2.json 且包含运动条目", + "key": "REQUIRE_MOBILITY_V2", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求对象存在 meta.json", + "key": "REQUIRE_META_JSON", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求 textured_objs 下存在真实 .obj mesh", + "key": "REQUIRE_TEXTURED_OBJS", + "type": "boolean", + "value": "true" + }, + { + "description": "是否要求对象存在 result.json", + "key": "REQUIRE_RESULT_JSON", + "type": "boolean", + "value": "false" + }, + { + "description": "是否要求 point_sample 下存在点云/标签文件", + "key": "REQUIRE_POINT_SAMPLE", + "type": "boolean", + "value": "false" + }, + { + "description": "是否要求对象存在 bounding_box.json", + "key": "REQUIRE_BOUNDING_BOX", + "type": "boolean", + "value": "false" + } + ], + "id": "operator-builtin-partnet-mobility-asset-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-partnet-mobility-asset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-partnet-mobility-asset-validate@sha256:52bb4ada32074e453f43aaf5fb9f7856a71eeb4e8ae6f2f0751943da55af93a0", + "inputs": [ + { + "description": "PartNet-Mobility 对象目录、dataset/sapien_dataset 根目录或 mobility.urdf 文件", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "PartNet-Mobility 资产校验", + "outputs": [ + { + "description": "PartNet-Mobility/SAPIEN 资产结构、关节树、运动语义和 mesh 引用校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_partnet_mobility_asset_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-partnet-mobility-asset-validate", + "operatorSlug": "partnet_mobility_asset_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "partnet-mobility-sapien-assets", + "urdf-files", + "json-files", + "semantics-txt", + "mesh-obj-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_partnet_mobility_asset_validate", + "runtimeSource": "backend/operators/builtin/partnet_mobility_asset_validate/operator.py", + "skillDefinition": "面向 SAPIEN/PartNet-Mobility、ManiSkill articulated object 和具身仿真数据生成资产入库,真实读取每个对象的 mobility.urdf、semantics.txt、mobility_v2.json、meta.json、result.json、result_after_merging.json、bounding_box.json、textured_objs/*.obj、point_sample 与 parts_render,校验 URDF link/joint 树、hinge/slider/screw/free 运动语义、motion limits/axis、mesh 引用和对象类别统计。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 MuJoCo/MJCF XML,校验 actuator 下 motor/position/velocity/general 等执行器目标引用、ctrlrange/forcerange/actrange 和 gear 数值。", + "hyperparams": [ + { + "description": "MJCF XML 文件或包含 MJCF 文件的目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多扫描多少个 MJCF 文件,必须为正整数", + "key": "MAX_MODELS", + "type": "string", + "value": "20" + } + ], + "id": "operator-builtin-mjcf-actuator-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-mjcf-actuator-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-mjcf-actuator-validate@sha256:adc49f2a4cd9a27f2e74c74852c8a74eb33233e98973e80507346260e2abed8c", + "inputs": [ + { + "description": "MJCF XML 文件或包含 MJCF 文件的目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "MJCF 执行器校验", + "outputs": [ + { + "description": "MJCF 执行器校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_mjcf_actuator_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-mjcf-actuator-validate", + "operatorSlug": "mjcf_actuator_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "mjcf-files", + "xml-parser" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_mjcf_actuator_validate", + "runtimeSource": "backend/operators/builtin/mjcf_actuator_validate/operator.py", + "skillDefinition": "面向 MuJoCo、robosuite、ManiSkill 和 GitHub 常见 MJCF 仿真资产,真实解析 下执行器,验证 joint/tendon/site/body 引用存在,ctrlrange/forcerange/actrange 为递增有限数值,并识别重复 actuator 名称。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 URDF/MJCF/XML 机器人模型,校验 link/joint/body/actuator 结构和 mesh/include 文件引用。", + "hyperparams": [ + { + "description": "URDF/MJCF/XML 文件或包含模型文件的目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "可选逗号分隔 package 根目录,用于解析 package:// mesh URI", + "key": "PACKAGE_ROOTS", + "type": "string", + "value": "" + }, + { + "description": "最多扫描多少个模型文件,必须为正整数", + "key": "MAX_MODELS", + "type": "string", + "value": "20" + }, + { + "description": "每个模型最多返回多少个 mesh/include 引用预览", + "key": "MAX_REFERENCES", + "type": "string", + "value": "100" + } + ], + "id": "operator-builtin-robot-model-asset-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-robot-model-asset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-robot-model-asset-validate@sha256:9e85004992d213c08d88dba1c4caf0221e56c6feee9e6e92ab48ce4b4c79fe93", + "inputs": [ + { + "description": "URDF/MJCF/XML 文件或机器人描述包目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "机器人模型资产校验", + "outputs": [ + { + "description": "机器人模型资产校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_robot_model_asset_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-robot-model-asset-validate", + "operatorSlug": "robot_model_asset_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "urdf-files", + "mjcf-files", + "mesh-assets" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_robot_model_asset_validate", + "runtimeSource": "backend/operators/builtin/robot_model_asset_validate/operator.py", + "skillDefinition": "面向 GitHub 常见 robot_description、URDF/MJCF 仿真资产和内部机器人模型入库,真实解析 XML,统计 link/joint/body/actuator,并解析 mesh filename/file、package:// 和 include 引用,识别缺失资产。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "解析真实 URDF joint limit,并读取 JSON/JSONL/CSV/Parquet 轨迹中的 qpos/joint_positions,校验关节位置是否越界。", + "hyperparams": [ + { + "description": "包含 URDF 与轨迹文件的目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "可选,显式 URDF 文件路径", + "key": "MODEL_PATH", + "type": "string", + "value": "" + }, + { + "description": "可选,显式 JSON/JSONL/CSV/Parquet 轨迹文件或目录", + "key": "TRAJECTORY_PATH", + "type": "string", + "value": "" + }, + { + "description": "逗号分隔的候选关节位置字段,支持点路径", + "key": "POSITION_FIELDS", + "type": "string", + "value": "joint_positions,observation.joint_positions,observation.state,state,states,qpos" + }, + { + "description": "逗号分隔的候选关节名字段,支持点路径", + "key": "JOINT_NAME_FIELDS", + "type": "string", + "value": "joint_names,observation.joint_names,names" + }, + { + "description": "可选,逗号分隔的轨迹列顺序;为空时使用 URDF 中有上下限的关节顺序", + "key": "JOINT_NAMES", + "type": "string", + "value": "" + }, + { + "description": "允许的限位浮点容差", + "key": "TOLERANCE", + "type": "string", + "value": "0" + }, + { + "description": "最多读取多少行真实轨迹", + "key": "MAX_ROWS", + "type": "string", + "value": "100000" + }, + { + "description": "最多读取多少个轨迹文件", + "key": "MAX_FILES", + "type": "string", + "value": "50" + } + ], + "id": "operator-builtin-joint-limit-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-joint-limit-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-joint-limit-validate@sha256:b4edf3d011f7234d3479635a6aecf33c695674827487d4c8578cb02446785b82", + "inputs": [ + { + "description": "包含 URDF 和轨迹文件的目录,或通过 MODEL_PATH/TRAJECTORY_PATH 指定真实文件", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "URDF 关节限位校验", + "outputs": [ + { + "description": "URDF 关节限位校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_joint_limit_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-joint-limit-validate", + "operatorSlug": "joint_limit_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "urdf-files", + "json-jsonl-csv-parquet-files", + "pyarrow" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_joint_limit_validate", + "runtimeSource": "backend/operators/builtin/joint_limit_validate/operator.py", + "skillDefinition": "面向真实机器人描述包和采集轨迹,解析 URDF revolute/prismatic joint 的 lower/upper limit,读取 joint_positions、observation.joint_positions、state 或 qpos 字段,按 JOINT_NAMES 或 URDF 顺序映射,发现维度不匹配和超限样本。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 PCD/PLY 点云文件,输出点数、字段、编码和 ASCII 点云空间边界。", + "hyperparams": [ + { + "description": ".pcd/.ply 文件或包含点云文件的目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多扫描多少个点云文件,必须为正整数", + "key": "MAX_FILES", + "type": "string", + "value": "50" + }, + { + "description": "每个 ASCII 点云最多采样多少点计算边界", + "key": "MAX_SAMPLE_POINTS", + "type": "string", + "value": "200000" + } + ], + "id": "operator-builtin-pointcloud-summary", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-pointcloud-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-pointcloud-summary@sha256:fa846da15ab5d4816d78375fc700873996fa0ca3a0aaeb9711f106e348146235", + "inputs": [ + { + "description": "PCD/PLY 文件或目录输入", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "点云数据概览", + "outputs": [ + { + "description": "点云文件概览", + "id": "out-1", + "name": "summary", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_pointcloud_summary", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-pointcloud-summary", + "operatorSlug": "pointcloud_summary", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "pcd-files", + "ply-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_pointcloud_summary", + "runtimeSource": "backend/operators/builtin/pointcloud_summary/operator.py", + "skillDefinition": "面向激光雷达、深度相机重建和具身空间感知数据,真实读取 PCD/PLY header、fields/properties、point count 和 ASCII x/y/z 采样边界,不把点云当普通文件名扫描。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 CSV/JSONL 时间戳清单,校验 camera/lidar/action 等多传感器流的同步完整性和最大漂移。", + "hyperparams": [ + { + "description": "CSV/JSONL 时间戳文件或包含时间戳文件的目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "逗号分隔的必需传感器流名称", + "key": "REQUIRED_STREAMS", + "type": "string", + "value": "camera,lidar,action" + }, + { + "description": "单个同步组允许的最大时间漂移毫秒数", + "key": "MAX_DRIFT_MS", + "type": "string", + "value": "50" + } + ], + "id": "operator-builtin-sensor-timestamp-sync-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-sensor-timestamp-sync-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-sensor-timestamp-sync-validate@sha256:4afd1efaf170d73917f215fbfdef55f36869726a32a07e79b9037204febe9a8b", + "inputs": [ + { + "description": "包含 stream/timestamp/frame_id 的 CSV 或 JSONL 时间戳文件", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "多传感器时间同步校验", + "outputs": [ + { + "description": "多传感器时间同步校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_sensor_timestamp_sync_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-sensor-timestamp-sync-validate", + "operatorSlug": "sensor_timestamp_sync_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "timestamp-csv-jsonl" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_sensor_timestamp_sync_validate", + "runtimeSource": "backend/operators/builtin/sensor_timestamp_sync_validate/operator.py", + "skillDefinition": "面向机器人多相机、LiDAR、action/state 采集链路,真实读取 timestamp/stamp、timestamp_ns/time_ns、timestamp_ms/time_ms、ROS sec+nsec 字段,按 frame_id/sequence 或最近邻分组,计算同步组缺失流和最大时间漂移。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 LeRobot v2/v3 数据集目录和 Parquet 数据文件,校验 meta、episode、schema、frame index 和视频引用。", + "hyperparams": [ + { + "description": "LeRobot 数据集根目录或其中任意文件路径", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "Parquet 必需列,逗号分隔", + "key": "REQUIRED_COLUMNS", + "type": "string", + "value": "episode_index,frame_index,timestamp,action" + }, + { + "description": "最多读取多少个 Parquet 文件的 schema 和索引", + "key": "MAX_PARQUET_FILES", + "type": "string", + "value": "24" + }, + { + "description": "最多扫描多少行 episode_index/frame_index 以检查顺序", + "key": "MAX_INDEX_ROWS", + "type": "string", + "value": "200000" + }, + { + "description": "是否检查每个 episode 内 frame_index 单调不降", + "key": "CHECK_INDEX_ORDER", + "type": "string", + "value": "true" + }, + { + "description": "当 info.json 声明 total_videos 时是否要求目录里存在真实视频文件", + "key": "CHECK_VIDEOS", + "type": "string", + "value": "true" + } + ], + "id": "operator-builtin-lerobot-dataset-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-dataset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-dataset-validate@sha256:f2471c2ea75c05059fd995667b337681f8f6ff3a8947c5b960f03fae93a2d087", + "inputs": [ + { + "description": "LeRobot 数据集根目录或上游目录 artifact", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "LeRobot 数据集校验", + "outputs": [ + { + "description": "LeRobot 数据集校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_lerobot_dataset_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-lerobot-dataset-validate", + "operatorSlug": "lerobot_dataset_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "lerobot-dataset-files", + "pyarrow", + "parquet-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_lerobot_dataset_validate", + "runtimeSource": "backend/operators/builtin/lerobot_dataset_validate/operator.py", + "skillDefinition": "面向 Hugging Face LeRobot 数据集真实落地验收,读取 meta/info.json、meta/episodes/*.parquet 或 episodes.jsonl、tasks.jsonl、stats.json 和 data/*.parquet,校验 schema 与 frame/episode 计数。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 LeRobot v3 data Parquet、meta/episodes 和 MP4 视频文件,校验声明视频数、视频帧数、数据行数、episode length 和显式视频引用是否一致。", + "hyperparams": [ + { + "description": "LeRobot 数据集根目录或其中任意文件路径", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多读取多少个 meta/data Parquet 文件", + "key": "MAX_PARQUET_FILES", + "type": "string", + "value": "50" + }, + { + "description": "最多扫描多少行真实 Parquet 数据以统计 episode 和视频引用", + "key": "MAX_ROWS", + "type": "string", + "value": "200000" + }, + { + "description": "最多读取多少个视频文件的帧数与基础属性", + "key": "MAX_VIDEOS", + "type": "string", + "value": "200" + }, + { + "description": "视频帧数与数据行数允许误差,必须为非负整数", + "key": "FRAME_TOLERANCE", + "type": "string", + "value": "0" + }, + { + "description": "是否要求存在真实视频文件", + "key": "REQUIRE_VIDEOS", + "type": "string", + "value": "true" + } + ], + "id": "operator-builtin-lerobot-video-alignment-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-video-alignment-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-video-alignment-validate@sha256:51022c000aed86bda84f3f33453984ed85620b87ea2487ed451864abcef7282a", + "inputs": [ + { + "description": "LeRobot 数据集根目录或上游目录 artifact", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "LeRobot 视频对齐校验", + "outputs": [ + { + "description": "LeRobot 视频对齐校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_lerobot_video_alignment_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-lerobot-video-alignment-validate", + "operatorSlug": "lerobot_video_alignment_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "lerobot-v3-data-files", + "pyarrow", + "parquet-files", + "video-files", + "opencv-python-headless" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_lerobot_video_alignment_validate", + "runtimeSource": "backend/operators/builtin/lerobot_video_alignment_validate/operator.py", + "skillDefinition": "面向 Hugging Face LeRobot v3 数据集、远程机器人视频轨迹和训练前数据落地验收,真实读取 meta/info.json 的 video feature、meta/episodes、data Parquet 和 videos 目录,使用 OpenCV 读取 MP4 帧数/FPS/尺寸,检查 Parquet 行数、episode 长度、视频文件数和视频帧数是否对齐。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 LeRobot v3 meta/info、stats、tasks、episodes 与 data parquet,校验声明计数、feature、stats 和 episode length 一致性。", + "hyperparams": [ + { + "description": "LeRobot 数据集根目录或其中任意文件路径", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "必需 feature/数据列,逗号分隔", + "key": "REQUIRED_FEATURES", + "type": "string", + "value": "action,observation.state" + }, + { + "description": "最多读取多少个 meta/data Parquet 文件", + "key": "MAX_PARQUET_FILES", + "type": "string", + "value": "50" + } + ], + "id": "operator-builtin-lerobot-metadata-consistency-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-metadata-consistency-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-metadata-consistency-validate@sha256:a58daedea8b3996999c73020a68a07f75c1bdbbcdb6189f5d8183c322a0d42c6", + "inputs": [ + { + "description": "LeRobot 数据集根目录或上游目录 artifact", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "LeRobot 元数据一致性校验", + "outputs": [ + { + "description": "LeRobot 元数据一致性校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_lerobot_metadata_consistency_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-lerobot-metadata-consistency-validate", + "operatorSlug": "lerobot_metadata_consistency_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "lerobot-v3-meta-files", + "pyarrow", + "parquet-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_lerobot_metadata_consistency_validate", + "runtimeSource": "backend/operators/builtin/lerobot_metadata_consistency_validate/operator.py", + "skillDefinition": "面向 Hugging Face LeRobot v3 数据集落地验收,真实读取 meta/info.json 的 features/total_*、meta/stats.json、meta/tasks.jsonl、meta/episodes Parquet/JSONL 与 data Parquet,检查 action/observation.state 等训练必需 feature 是否声明、是否有 stats、是否真的出现在数据列中,以及 episode length 和 task 索引是否一致。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 LeRobot v3 meta/episodes 与 data Parquet,校验 episode metadata 的 length、offset、chunk/data path template 与实际 data shard 行范围是否一致。", + "hyperparams": [ + { + "description": "LeRobot v3 数据集根目录或其中任意文件路径", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多扫描多少个 episode/data Parquet 文件", + "key": "MAX_PARQUET_FILES", + "type": "string", + "value": "100" + }, + { + "description": "未显式提供 chunk_index 时用于 episode_index 推导 chunk 的大小", + "key": "CHUNK_SIZE", + "type": "string", + "value": "1000" + }, + { + "description": "可选 data path 模板,例如 data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet", + "key": "DATA_PATH_TEMPLATE", + "type": "string", + "value": "" + }, + { + "description": "候选 data path 存在时是否要求该文件实际包含对应 episode 行", + "key": "STRICT_PATH_MATCH", + "type": "string", + "value": "true" + } + ], + "id": "operator-builtin-lerobot-v3-episode-index-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-v3-episode-index-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-v3-episode-index-validate@sha256:204d8dc1e1530c8096c7aca6817ce41b12ed2c70fcd0d31fe2346427b0833456", + "inputs": [ + { + "description": "LeRobot v3 数据集根目录或上游目录 artifact", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "LeRobot v3 Episode索引校验", + "outputs": [ + { + "description": "LeRobot v3 episode 索引校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_lerobot_v3_episode_index_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-lerobot-v3-episode-index-validate", + "operatorSlug": "lerobot_v3_episode_index_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "lerobot-v3-meta-data-files", + "pyarrow", + "parquet-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_lerobot_v3_episode_index_validate", + "runtimeSource": "backend/operators/builtin/lerobot_v3_episode_index_validate/operator.py", + "skillDefinition": "面向 Hugging Face LeRobot v3 大规模具身数据集,真实读取 meta/info.json、meta/episodes/*.parquet 和 data/**/*.parquet,验证 data_path template、chunk_index、episode length、data row window 与实际 Parquet shard 映射是否一致,补齐训练前对 episode 索引错位和 shard 漏行的验收。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 LeRobot 数据集目录,输出 meta/tasks/parquet/videos 概览。", + "hyperparams": [ + { + "description": "LeRobot 数据集根目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + } + ], + "id": "operator-builtin-lerobot-summary", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-summary@sha256:c8e1878dced01da3da3a068cd53a395b03061b27f3e52e20cef2faedf4081dbf", + "inputs": [ + { + "description": "目录输入", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "LeRobot 数据集概览", + "outputs": [ + { + "description": "概览结果", + "id": "out-1", + "name": "summary", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_lerobot_summary", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-lerobot-summary", + "operatorSlug": "lerobot_summary", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "lerobot-dataset-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_lerobot_summary", + "runtimeSource": "backend/operators/builtin/lerobot_summary/operator.py", + "skillDefinition": "面向 LeRobot 数据集目录的真实概览算子,适合机器人示教数据检查。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 rosbag1 / rosbag2 / mcap / db3 数据并输出概览。", + "hyperparams": [ + { + "description": "bag/db3/mcap 文件或 rosbag2 目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + } + ], + "id": "operator-builtin-rosbag-summary", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rosbag-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rosbag-summary@sha256:9d41ed0073c84068869b7c844cf2a5362601dab551835a0ce0a3b1a9d00de9c5", + "inputs": [ + { + "description": "bag 输入", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "ROSBag 数据概览", + "outputs": [ + { + "description": "概览结果", + "id": "out-1", + "name": "summary", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_rosbag_summary", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-rosbag-summary", + "operatorSlug": "rosbag_summary", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "rosbag2-files", + "sqlite" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_rosbag_summary", + "runtimeSource": "backend/operators/builtin/rosbag_summary/operator.py", + "skillDefinition": "支持 rosbag db3 topic/message 概览,以及 bag/mcap 文件摘要。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 MCAP 文件,解析 schema、channel、topic 和 message payload 统计。", + "hyperparams": [ + { + "description": ".mcap 文件或包含 .mcap 文件的目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多扫描多少条 message,必须为正整数", + "key": "MAX_MESSAGES", + "type": "string", + "value": "200000" + } + ], + "id": "operator-builtin-mcap-topic-summary", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-mcap-topic-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-mcap-topic-summary@sha256:421e060f3d74ac067b895d83a4c7478c43feca5a75d91623859b5a1d867a5ecc", + "inputs": [ + { + "description": "MCAP 文件或目录输入", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "MCAP 主题概览", + "outputs": [ + { + "description": "MCAP topic/schema/message 概览", + "id": "out-1", + "name": "summary", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_mcap_topic_summary", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-mcap-topic-summary", + "operatorSlug": "mcap_topic_summary", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "mcap-files", + "mcap-python" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_mcap_topic_summary", + "runtimeSource": "backend/operators/builtin/mcap_topic_summary/operator.py", + "skillDefinition": "面向 ROS2/机器人传感器日志的真实 MCAP 数据,使用 mcap runtime 解析 topic、schema、channel、message payload bytes 和时间范围,而不是只读取文件头。", + "type": "builtin" + }, + { + "category": "ROBOTICS", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 TFRecord/RLDS shard,校验 record framing、长度、CRC32C,并输出样本预览。", + "hyperparams": [ + { + "description": ".tfrecord/.tfrecords/.tfrec 文件或包含这些文件的目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "最多扫描多少条 record,必须为正整数", + "key": "MAX_RECORDS", + "type": "string", + "value": "200000" + }, + { + "description": "最多返回多少条 record 预览,必须为正整数", + "key": "PREVIEW_RECORDS", + "type": "string", + "value": "5" + }, + { + "description": "是否校验 TFRecord masked CRC32C", + "key": "VALIDATE_CRC", + "type": "string", + "value": "true" + } + ], + "id": "operator-builtin-tfrecord-structural-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-tfrecord-structural-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-tfrecord-structural-validate@sha256:60fb1eb05a35a6b7712c5e07ce067962ed125ef3f3e8b73853c56c5d250e30ad", + "inputs": [ + { + "description": "TFRecord 文件或目录输入", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "TFRecord 结构校验", + "outputs": [ + { + "description": "TFRecord 结构校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_tfrecord_structural_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-tfrecord-structural-validate", + "operatorSlug": "tfrecord_structural_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "tfrecord-files", + "crc32c-structural-validation" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_tfrecord_structural_validate", + "runtimeSource": "backend/operators/builtin/tfrecord_structural_validate/operator.py", + "skillDefinition": "面向 RLDS/Open X-Embodiment 等 TensorFlow 数据链路,真实读取 TFRecord 二进制 record framing,校验 length/header CRC 和 payload CRC,识别截断或损坏 shard。", + "type": "builtin" + }, + { + "category": "DATASET", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 CSV/JSON/JSONL/Parquet 切分清单,校验 train/val/test 等 split 之间是否存在 episode 泄漏、缺失 split 或重复样本。", + "hyperparams": [ + { + "description": "真实 split manifest 文件或包含 split 清单的数据集目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "必须存在的 split 名称,逗号分隔;可按业务改为 train,val 或其它命名", + "key": "REQUIRED_SPLITS", + "type": "string", + "value": "train,val,test" + }, + { + "description": "每个必需 split 的最小样本数", + "key": "MIN_ROWS_PER_SPLIT", + "type": "number", + "value": "1" + }, + { + "description": "候选 episode ID 字段,逗号分隔", + "key": "EPISODE_ID_FIELDS", + "type": "string", + "value": "episodeId,episode_id,episode_index,episodeIndex,trajectory_id,trajectoryId,demoId,id,name" + }, + { + "description": "候选 split 字段,逗号分隔", + "key": "SPLIT_FIELDS", + "type": "string", + "value": "split,subset,set,fold" + }, + { + "description": "候选任务字段,用于统计任务在各 split 的分布", + "key": "TASK_FIELDS", + "type": "string", + "value": "task,taskName,task_index,taskIndex,language_instruction,instruction" + }, + { + "description": "最多扫描的清单文件数", + "key": "MAX_FILES", + "type": "number", + "value": "50" + }, + { + "description": "最多扫描的清单行数", + "key": "MAX_ROWS", + "type": "number", + "value": "200000" + } + ], + "id": "operator-builtin-dataset-split-leakage-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dataset-split-leakage-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dataset-split-leakage-validate@sha256:2dde8d42ac5001e4727b6061c54970c136f7588c1bde04fa7b5e3473e49aaac4", + "inputs": [ + { + "description": "可选上游数据;通常使用 SOURCE_PATH 指向真实切分清单或数据集目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "数据集切分泄漏校验", + "outputs": [ + { + "description": "切分泄漏校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_dataset_split_leakage_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-dataset-split-leakage-validate", + "operatorSlug": "dataset_split_leakage_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "split-manifest-csv-json-jsonl-parquet-files", + "pyarrow" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_dataset_split_leakage_validate", + "runtimeSource": "backend/operators/builtin/dataset_split_leakage_validate/operator.py", + "skillDefinition": "面向具身训练集发布和验收,确保 episode 不会同时出现在 train/val/test 等不同 split 中,并统计每个任务在各 split 的覆盖情况。", + "type": "builtin" + }, + { + "category": "DATASET", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 CSV/JSON/JSONL/Parquet manifest 中的图片、视频、深度图、mask、点云引用,检查文件存在并可被对应解码器读取。", + "hyperparams": [ + { + "description": "真实 manifest 文件或包含 manifest 与媒体文件的数据集目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "显式媒体引用字段,支持点路径,逗号分隔", + "key": "MEDIA_FIELDS", + "type": "string", + "value": "image,image_path,imagePath,video,video_path,videoPath,rgb,rgb_path,depth,depth_path,mask,mask_path,pointcloud,pointcloud_path,observation.image,observation.rgb,observation.depth,observation.video,observation.images.rgb,observation.images.wrist" + }, + { + "description": "是否额外从字段名包含 image/video/rgb/depth/mask/path 等关键词的列中推断媒体引用", + "key": "INFER_FIELDS", + "type": "boolean", + "value": "true" + }, + { + "description": "是否读取并解码图片、深度数组、视频元数据或点云文件头", + "key": "DEEP_CHECK", + "type": "boolean", + "value": "true" + }, + { + "description": "最多扫描的 manifest 文件数", + "key": "MAX_FILES", + "type": "number", + "value": "50" + }, + { + "description": "最多扫描的 manifest 行数", + "key": "MAX_ROWS", + "type": "number", + "value": "200000" + }, + { + "description": "最多检查的媒体引用数量", + "key": "MAX_REFERENCES", + "type": "number", + "value": "5000" + } + ], + "id": "operator-builtin-media-reference-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-media-reference-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-media-reference-validate@sha256:60bdd6320deaf66576927bf26b25809911ea98210e9c2daa0665fe9846157104", + "inputs": [ + { + "description": "可选上游数据;通常使用 SOURCE_PATH 指向真实 manifest 或数据集目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "媒体引用可读性校验", + "outputs": [ + { + "description": "媒体引用校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_media_reference_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-media-reference-validate", + "operatorSlug": "media_reference_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "csv-json-jsonl-parquet-manifests", + "image-video-depth-pointcloud-files", + "opencv-python-headless", + "numpy", + "pyarrow", + "ffprobe" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_media_reference_validate", + "runtimeSource": "backend/operators/builtin/media_reference_validate/operator.py", + "skillDefinition": "面向 HuggingFace/GitHub 常见 manifest + media 文件组织,防止训练前才发现图片、视频、深度帧或点云路径缺失、不可读。", + "type": "builtin" + }, + { + "category": "DATASET", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "读取真实 CSV/JSON/JSONL/Parquet/HDF5/NPZ 文件,按必需列、HDF5 dataset、NPZ array 合约校验训练输入 schema。", + "hyperparams": [ + { + "description": "真实 CSV/JSON/JSONL/Parquet/HDF5/NPZ 文件或目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "表格类文件必须存在的列或字段,逗号分隔;可用 CONTRACT_JSON.requiredColumns 替代", + "key": "REQUIRED_COLUMNS", + "type": "string", + "value": "" + }, + { + "description": "HDF5 文件必须存在的 dataset 路径,逗号分隔;可用 CONTRACT_JSON.requiredHdf5Datasets 替代", + "key": "REQUIRED_HDF5_DATASETS", + "type": "string", + "value": "" + }, + { + "description": "NPZ 文件必须存在的 array 名称,逗号分隔;可用 CONTRACT_JSON.requiredNpzArrays 替代", + "key": "REQUIRED_NPZ_ARRAYS", + "type": "string", + "value": "" + }, + { + "description": "可选 JSON 合约对象,支持 requiredColumns、requiredHdf5Datasets、requiredNpzArrays", + "key": "CONTRACT_JSON", + "type": "json", + "value": "" + }, + { + "description": "最多扫描的数据文件数", + "key": "MAX_FILES", + "type": "number", + "value": "50" + }, + { + "description": "表格类文件最多采样行数", + "key": "MAX_ROWS", + "type": "number", + "value": "10000" + } + ], + "id": "operator-builtin-dataset-schema-contract-validate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dataset-schema-contract-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dataset-schema-contract-validate@sha256:9dfcb8cf18e09069e14db67c8649352071fcdf44b1d537465d38aff532f5ba48", + "inputs": [ + { + "description": "可选上游数据;通常使用 SOURCE_PATH 指向真实数据文件或目录", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "数据集Schema合约校验", + "outputs": [ + { + "description": "Schema 合约校验结果", + "id": "out-1", + "name": "validation", + "transmission": "bus", + "type": "json" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_dataset_schema_contract_validate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-dataset-schema-contract-validate", + "operatorSlug": "dataset_schema_contract_validate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "csv-json-jsonl-parquet-hdf5-npz-files", + "pyarrow", + "h5py", + "numpy" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_dataset_schema_contract_validate", + "runtimeSource": "backend/operators/builtin/dataset_schema_contract_validate/operator.py", + "skillDefinition": "把具身训练前的 schema 假设显式化,统一检查表格列、HDF5 dataset 与 NPZ array 是否存在且非空。", + "type": "builtin" + }, + { + "category": "AI", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "description": "基于真实 VLA 样本清单训练轻量线性策略模型,作为 CPU 可运行的具身策略基线。", + "hyperparams": [ + { + "description": "可选样本清单路径,支持 JSON/JSONL/CSV 或目录", + "key": "SOURCE_PATH", + "type": "string", + "value": "" + }, + { + "description": "输出模型名称", + "key": "MODEL_NAME", + "type": "string", + "value": "vla-linear-policy" + }, + { + "description": "语言指令字段候选,逗号分隔", + "key": "TEXT_FIELDS", + "type": "string", + "value": "" + }, + { + "description": "动作字段候选,逗号分隔", + "key": "ACTION_FIELDS", + "type": "string", + "value": "" + }, + { + "description": "最少训练样本数", + "key": "MIN_SAMPLES", + "type": "number", + "value": "2" + } + ], + "id": "operator-builtin-vla-model-train", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-vla-model-train:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-vla-model-train@sha256:eccc4d24b9a8f0a69d9de173f3eb7b4541550245138babdd68571a77cd91fc45", + "inputs": [ + { + "description": "包含 instruction 与 action 的 VLA 样本清单或上游样本 artifact", + "id": "in-1", + "name": "samples", + "transmission": "bus", + "type": "json" + } + ], + "name": "轻量线性策略训练", + "outputs": [ + { + "description": "训练后的 VLA 策略模型与训练指标", + "id": "out-1", + "name": "model", + "transmission": "bus", + "type": "json" } ], - "runtimeSource": "backend/operators/builtin/vla_model_train/operator.py", - "runtimeHandler": "builtin_vla_model_train", "resources": { "cpuLimit": "2", "memoryLimit": "4Gi" - } + }, + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_vla_model_train", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-vla-model-train", + "operatorSlug": "vla_model_train", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "vla-sample-action-manifest", + "json-jsonl-csv-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_vla_model_train", + "runtimeSource": "backend/operators/builtin/vla_model_train/operator.py", + "skillDefinition": "读取真实 VLA 样本并拟合可复现的线性策略基线;缺少真实 action 时失败。该算子不冒充 ACT/OpenVLA 等深度模型训练。", + "type": "builtin" }, { - "id": "operator-builtin-act-policy-train", - "name": "ACT 策略训练", - "type": "builtin", "category": "AI", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "description": "在 A800 上运行真实 ACT imitation learning 训练命令,面向 ALOHA/LeRobot/HDF5 示教数据。", - "inputs": [ - { - "id": "in-1", - "name": "dataset", - "type": "json", - "description": "训练数据 manifest、切分清单或上游数据 artifact", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "model", - "type": "file", - "description": "训练得到的模型、权重、指标和日志 artifact", - "transmission": "bus" - } - ], - "skillDefinition": "具体 ACT 策略训练算子;平台负责 GPU 调度、上下游 artifact 注入和结果收集,训练本体必须由 COMMAND/SCRIPT 指向真实 ACT/LeRobot/自研训练代码,不接受无命令空跑。推荐 1 张 A800 80GB。", "hyperparams": [ { - "key": "COMMAND", - "value": "", - "description": "真实训练/推理命令;容器内执行,必须读取上游 INPUT_ARTIFACTS_PATH 或 SOURCE_PATH 并写出模型/报告文件", - "type": "string" + "description": "真实训练数据路径;支持 JSON/JSONL/CSV 或包含 samples.json/jsonl 的目录", + "key": "DATASET_PATH", + "type": "string", + "value": "" }, { - "key": "SCRIPT", - "value": "", - "description": "可选脚本内容;与 COMMAND 二选一", - "type": "string" + "description": "状态向量字段名", + "key": "STATE_FIELD", + "type": "string", + "value": "state" }, { - "key": "SOURCE_PATH", - "value": "", - "description": "可选真实数据集、模型或 manifest 路径", - "type": "string" + "description": "动作向量字段名", + "key": "ACTION_FIELD", + "type": "string", + "value": "action" }, { - "key": "OUTPUT_MODEL_NAME", - "value": "", - "description": "输出模型或权重名称", - "type": "string" + "description": "训练轮数", + "key": "EPOCHS", + "type": "number", + "value": "5" }, { - "key": "MODEL_FAMILY", - "value": "ACT", - "description": "模型族", - "type": "string" + "description": "训练 batch size", + "key": "BATCH_SIZE", + "type": "number", + "value": "8" + }, + { + "description": "AdamW 学习率", + "key": "LEARNING_RATE", + "type": "number", + "value": "0.001" + }, + { + "description": "策略网络隐藏层宽度", + "key": "HIDDEN_DIM", + "type": "number", + "value": "128" + }, + { + "description": "ACT action chunk size", + "key": "CHUNK_SIZE", + "type": "number", + "value": "1" }, { - "key": "RECOMMENDED_GPU", - "value": "1x NVIDIA A800 80GB", "description": "推荐 GPU 配置", - "type": "string" + "key": "RECOMMENDED_GPU", + "type": "string", + "value": "1x NVIDIA A800 80GB" + } + ], + "id": "operator-builtin-act-policy-train", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-act-policy-train:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-act-policy-train@sha256:d24b5417b2eab1abf5bb07a1d66d38306ea5545226d3f2a3e34bb5aee184f4f2", + "inputs": [ + { + "description": "训练数据 manifest、切分清单或上游数据 artifact", + "id": "in-1", + "name": "dataset", + "transmission": "bus", + "type": "json" + } + ], + "name": "ACT 策略训练", + "outputs": [ + { + "description": "训练得到的模型、权重、指标和日志 artifact", + "id": "out-1", + "name": "model", + "transmission": "bus", + "type": "file" } ], "resources": { "cpuLimit": "16", - "memoryLimit": "64Gi", "gpuLimit": "1", - "gpuType": "NVIDIA A800 80GB" + "gpuType": "NVIDIA A800 80GB", + "memoryLimit": "64Gi" }, + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": true, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_act_policy_train", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-act-policy-train", + "operatorSlug": "act_policy_train", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 1, + "recommendedGpuType": "NVIDIA A800 80GB", + "requiresGpu": true + }, + "runtimeDependencies": [ + "real-act-training-command", + "trajectory-demonstration-files", + "gpu-a800" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_act_policy_train", "runtimeSource": "backend/operators/builtin/act_policy_train/operator.py", - "runtimeHandler": "builtin_act_policy_train" + "skillDefinition": "ACT 策略训练算子;使用 DATASET_PATH/STATE_FIELD/ACTION_FIELD 等结构化参数读取真实轨迹样本,在独立 GPU 镜像中训练并输出 checkpoint.pt 与 metrics.json。推荐 1 张 A800 80GB。", + "type": "builtin" }, { - "id": "operator-builtin-diffusion-policy-train", - "name": "Diffusion Policy 训练", - "type": "builtin", "category": "AI", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "description": "在 A800 上运行真实 Diffusion Policy 训练命令,面向 UMI/Zarr/LeRobot 等轨迹数据。", - "inputs": [ - { - "id": "in-1", - "name": "dataset", - "type": "json", - "description": "训练数据 manifest、切分清单或上游数据 artifact", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "model", - "type": "file", - "description": "训练得到的模型、权重、指标和日志 artifact", - "transmission": "bus" - } - ], - "skillDefinition": "具体 Diffusion Policy 训练算子;平台提供 GPU Job、artifact 和血缘,训练命令必须读取真实轨迹数据并产出 checkpoint/metrics。推荐 1 张 A800 80GB。", "hyperparams": [ { - "key": "COMMAND", - "value": "", - "description": "真实训练/推理命令;容器内执行,必须读取上游 INPUT_ARTIFACTS_PATH 或 SOURCE_PATH 并写出模型/报告文件", - "type": "string" + "description": "真实训练数据路径;支持 JSON/JSONL/CSV 或包含 samples.json/jsonl 的目录", + "key": "DATASET_PATH", + "type": "string", + "value": "" }, { - "key": "SCRIPT", - "value": "", - "description": "可选脚本内容;与 COMMAND 二选一", - "type": "string" + "description": "观测/状态向量字段名", + "key": "STATE_FIELD", + "type": "string", + "value": "state" }, { - "key": "SOURCE_PATH", - "value": "", - "description": "可选真实数据集、模型或 manifest 路径", - "type": "string" + "description": "动作向量字段名", + "key": "ACTION_FIELD", + "type": "string", + "value": "action" }, { - "key": "OUTPUT_MODEL_NAME", - "value": "", - "description": "输出模型或权重名称", - "type": "string" + "description": "训练轮数", + "key": "EPOCHS", + "type": "number", + "value": "5" }, { - "key": "MODEL_FAMILY", - "value": "Diffusion Policy", - "description": "模型族", - "type": "string" + "description": "训练 batch size", + "key": "BATCH_SIZE", + "type": "number", + "value": "8" + }, + { + "description": "AdamW 学习率", + "key": "LEARNING_RATE", + "type": "number", + "value": "0.001" + }, + { + "description": "策略网络隐藏层宽度", + "key": "HIDDEN_DIM", + "type": "number", + "value": "256" + }, + { + "description": "预测 horizon", + "key": "HORIZON", + "type": "number", + "value": "16" + }, + { + "description": "扩散步数配置,用于训练记录和下游复现", + "key": "DIFFUSION_STEPS", + "type": "number", + "value": "32" }, { - "key": "RECOMMENDED_GPU", - "value": "1x NVIDIA A800 80GB", "description": "推荐 GPU 配置", - "type": "string" + "key": "RECOMMENDED_GPU", + "type": "string", + "value": "1x NVIDIA A800 80GB" + } + ], + "id": "operator-builtin-diffusion-policy-train", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-diffusion-policy-train:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-diffusion-policy-train@sha256:78d74d404ed7e98c19518e28315a897313c8583c6f6b0eb191e36eabcc8a2f33", + "inputs": [ + { + "description": "训练数据 manifest、切分清单或上游数据 artifact", + "id": "in-1", + "name": "dataset", + "transmission": "bus", + "type": "json" + } + ], + "name": "Diffusion Policy 训练", + "outputs": [ + { + "description": "训练得到的模型、权重、指标和日志 artifact", + "id": "out-1", + "name": "model", + "transmission": "bus", + "type": "file" } ], "resources": { "cpuLimit": "16", - "memoryLimit": "96Gi", "gpuLimit": "1", - "gpuType": "NVIDIA A800 80GB" + "gpuType": "NVIDIA A800 80GB", + "memoryLimit": "96Gi" }, + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": true, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_diffusion_policy_train", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-diffusion-policy-train", + "operatorSlug": "diffusion_policy_train", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 1, + "recommendedGpuType": "NVIDIA A800 80GB", + "requiresGpu": true + }, + "runtimeDependencies": [ + "diffusion-policy-training-command", + "zarr-or-lerobot-trajectory-files", + "gpu-a800" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_diffusion_policy_train", "runtimeSource": "backend/operators/builtin/diffusion_policy_train/operator.py", - "runtimeHandler": "builtin_diffusion_policy_train" + "skillDefinition": "Diffusion Policy 训练算子;使用 DATASET_PATH/STATE_FIELD/ACTION_FIELD/HORIZON/DIFFUSION_STEPS 等结构化参数读取真实轨迹数据,输出 checkpoint.pt 与 metrics.json。推荐 1 张 A800 80GB。", + "type": "builtin" }, { - "id": "operator-builtin-openvla-finetune", - "name": "OpenVLA 微调训练", - "type": "builtin", "category": "AI", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "description": "在 A800 上运行真实 OpenVLA LoRA/QLoRA 微调命令,面向视觉语言动作数据。", - "inputs": [ - { - "id": "in-1", - "name": "dataset", - "type": "json", - "description": "训练数据 manifest、切分清单或上游数据 artifact", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "model", - "type": "file", - "description": "训练得到的模型、权重、指标和日志 artifact", - "transmission": "bus" - } - ], - "skillDefinition": "具体 OpenVLA 微调算子;推荐先用 LoRA/QLoRA 在 1 张 A800 80GB 上训练,full fine-tune 可在命令中改为多卡。必须提供真实权重、数据和训练命令。", "hyperparams": [ { - "key": "COMMAND", - "value": "", - "description": "真实训练/推理命令;容器内执行,必须读取上游 INPUT_ARTIFACTS_PATH 或 SOURCE_PATH 并写出模型/报告文件", - "type": "string" - }, - { - "key": "SCRIPT", - "value": "", - "description": "可选脚本内容;与 COMMAND 二选一", - "type": "string" - }, - { - "key": "SOURCE_PATH", - "value": "", - "description": "可选真实数据集、模型或 manifest 路径", - "type": "string" - }, - { - "key": "OUTPUT_MODEL_NAME", - "value": "", - "description": "输出模型或权重名称", - "type": "string" - }, - { + "description": "Hugging Face 或本地基础模型路径", "key": "BASE_MODEL", - "value": "openvla/openvla-7b", - "description": "基础模型或本地权重路径", - "type": "string" + "type": "string", + "value": "" }, { - "key": "FINETUNE_METHOD", - "value": "lora", - "description": "lora、qlora 或 full", - "type": "string" + "description": "真实微调数据 JSON/JSONL/CSV 路径", + "key": "DATASET_PATH", + "type": "string", + "value": "" + }, + { + "description": "指令/观测文本字段名", + "key": "TEXT_FIELD", + "type": "string", + "value": "instruction" + }, + { + "description": "目标动作文本字段名", + "key": "TARGET_FIELD", + "type": "string", + "value": "action" + }, + { + "description": "token 最大长度", + "key": "MAX_LENGTH", + "type": "number", + "value": "256" + }, + { + "description": "训练轮数", + "key": "EPOCHS", + "type": "number", + "value": "1" + }, + { + "description": "训练 batch size", + "key": "BATCH_SIZE", + "type": "number", + "value": "1" + }, + { + "description": "AdamW 学习率", + "key": "LEARNING_RATE", + "type": "number", + "value": "0.00005" + }, + { + "description": "推荐精度配置", + "key": "PRECISION", + "type": "string", + "value": "bf16" }, { - "key": "RECOMMENDED_GPU", - "value": "1x NVIDIA A800 80GB", "description": "推荐 GPU 配置", - "type": "string" + "key": "RECOMMENDED_GPU", + "type": "string", + "value": "1x NVIDIA A800 80GB" + } + ], + "id": "operator-builtin-openvla-finetune", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-openvla-finetune:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-openvla-finetune@sha256:36184d088fb2ca79a69c6c11adfba4099a1afa32eac7ec3da6408946044837d2", + "inputs": [ + { + "description": "训练数据 manifest、切分清单或上游数据 artifact", + "id": "in-1", + "name": "dataset", + "transmission": "bus", + "type": "json" + } + ], + "name": "OpenVLA 微调训练", + "outputs": [ + { + "description": "训练得到的模型、权重、指标和日志 artifact", + "id": "out-1", + "name": "model", + "transmission": "bus", + "type": "file" } ], "resources": { "cpuLimit": "24", - "memoryLimit": "128Gi", "gpuLimit": "1", - "gpuType": "NVIDIA A800 80GB" + "gpuType": "NVIDIA A800 80GB", + "memoryLimit": "128Gi" }, + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": true, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_openvla_finetune", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-openvla-finetune", + "operatorSlug": "openvla_finetune", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 1, + "recommendedGpuType": "NVIDIA A800 80GB", + "requiresGpu": true + }, + "runtimeDependencies": [ + "openvla-training-command", + "vla-sample-action-manifest", + "gpu-a800" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_openvla_finetune", "runtimeSource": "backend/operators/builtin/openvla_finetune/operator.py", - "runtimeHandler": "builtin_openvla_finetune" + "skillDefinition": "OpenVLA 微调算子;使用 BASE_MODEL、DATASET_PATH、TEXT_FIELD、TARGET_FIELD 等结构化参数,通过 Transformers/PyTorch 对真实指令-动作样本训练并输出模型目录与 metrics.json。推荐 1 张 A800 80GB。", + "type": "builtin" }, { - "id": "operator-builtin-openvla-inference", - "name": "OpenVLA 推理", - "type": "builtin", "category": "AI", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "description": "在 A800 上运行真实 OpenVLA/视觉语言动作模型推理命令,输出动作预测。", - "inputs": [ - { - "id": "in-1", - "name": "model-and-samples", - "type": "json", - "description": "模型权重、样本清单或上游 artifact", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "predictions", - "type": "json", - "description": "动作预测、置信度和推理日志", - "transmission": "bus" - } - ], - "skillDefinition": "具体 OpenVLA 推理算子;用于替代笼统模型推理,必须由 COMMAND/SCRIPT 调用真实模型权重并生成预测 artifact。推荐 1 张 A800,量化模型可降低显存。", "hyperparams": [ { - "key": "COMMAND", - "value": "", - "description": "真实训练/推理命令;容器内执行,必须读取上游 INPUT_ARTIFACTS_PATH 或 SOURCE_PATH 并写出模型/报告文件", - "type": "string" - }, - { - "key": "SCRIPT", - "value": "", - "description": "可选脚本内容;与 COMMAND 二选一", - "type": "string" - }, - { - "key": "SOURCE_PATH", - "value": "", - "description": "可选真实数据集、模型或 manifest 路径", - "type": "string" - }, - { - "key": "OUTPUT_MODEL_NAME", - "value": "", - "description": "输出模型或权重名称", - "type": "string" - }, - { + "description": "本地模型目录或 Hugging Face 模型名", "key": "MODEL_PATH", - "value": "", - "description": "OpenVLA 权重或 adapter 路径", - "type": "string" + "type": "string", + "value": "" }, { + "description": "未提供 MODEL_PATH 时使用的基础模型", + "key": "BASE_MODEL", + "type": "string", + "value": "" + }, + { + "description": "推理 prompt;为空则从上游样本读取 TEXT_FIELD", + "key": "PROMPT", + "type": "string", + "value": "" + }, + { + "description": "上游样本指令字段名", + "key": "TEXT_FIELD", + "type": "string", + "value": "instruction" + }, + { + "description": "生成 token 数", + "key": "MAX_NEW_TOKENS", + "type": "number", + "value": "64" + }, + { + "description": "推荐精度配置", "key": "PRECISION", - "value": "bf16", - "description": "bf16、fp16、int8 或 int4", - "type": "string" + "type": "string", + "value": "bf16" }, { - "key": "RECOMMENDED_GPU", - "value": "1x NVIDIA A800 80GB", "description": "推荐 GPU 配置", - "type": "string" + "key": "RECOMMENDED_GPU", + "type": "string", + "value": "1x NVIDIA A800 80GB" + } + ], + "id": "operator-builtin-openvla-inference", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-openvla-inference:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-openvla-inference@sha256:3336a89556d87b18a268be0234ff95acd6b23405e90c0da6fe9cdcbf7d3c524e", + "inputs": [ + { + "description": "模型权重、样本清单或上游 artifact", + "id": "in-1", + "name": "model-and-samples", + "transmission": "bus", + "type": "json" + } + ], + "name": "OpenVLA 推理", + "outputs": [ + { + "description": "动作预测、置信度和推理日志", + "id": "out-1", + "name": "predictions", + "transmission": "bus", + "type": "json" } ], "resources": { "cpuLimit": "12", - "memoryLimit": "64Gi", "gpuLimit": "1", - "gpuType": "NVIDIA A800 80GB" + "gpuType": "NVIDIA A800 80GB", + "memoryLimit": "64Gi" }, + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": true, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_openvla_inference", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-openvla-inference", + "operatorSlug": "openvla_inference", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 1, + "recommendedGpuType": "NVIDIA A800 80GB", + "requiresGpu": true + }, + "runtimeDependencies": [ + "openvla-inference-command", + "openvla-model-weights", + "gpu-a800" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_openvla_inference", "runtimeSource": "backend/operators/builtin/openvla_inference/operator.py", - "runtimeHandler": "builtin_openvla_inference" + "skillDefinition": "OpenVLA 推理算子;使用 MODEL_PATH 或 BASE_MODEL 加载真实 Transformers 权重,读取 PROMPT 或上游样本 instruction,并输出 inference-result.json。推荐 1 张 A800,量化模型可降低显存。", + "type": "builtin" }, { - "id": "operator-builtin-yolo-embodied-detection-train", - "name": "YOLO 具身目标检测训练", - "type": "builtin", "category": "AI", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "description": "在 A800 上运行真实 YOLO 检测训练命令,面向机器人视频、抓取物体和合成视觉数据。", - "inputs": [ - { - "id": "in-1", - "name": "dataset", - "type": "json", - "description": "训练数据 manifest、切分清单或上游数据 artifact", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "model", - "type": "file", - "description": "训练得到的模型、权重、指标和日志 artifact", - "transmission": "bus" - } - ], - "skillDefinition": "具体 YOLO 检测训练算子;必须读取真实 YOLO/COCO 数据集并输出权重、metrics 和类别配置。A800 对 YOLOv8/YOLOv10/YOLO11 中大模型训练充足,默认请求 1 卡。", "hyperparams": [ { - "key": "COMMAND", - "value": "", - "description": "真实训练/推理命令;容器内执行,必须读取上游 INPUT_ARTIFACTS_PATH 或 SOURCE_PATH 并写出模型/报告文件", - "type": "string" - }, - { - "key": "SCRIPT", - "value": "", - "description": "可选脚本内容;与 COMMAND 二选一", - "type": "string" - }, - { - "key": "SOURCE_PATH", - "value": "", - "description": "可选真实数据集、模型或 manifest 路径", - "type": "string" - }, - { - "key": "OUTPUT_MODEL_NAME", - "value": "", - "description": "输出模型或权重名称", - "type": "string" - }, - { - "key": "MODEL_SIZE", - "value": "yolo11m", - "description": "建议模型规格", - "type": "string" - }, - { + "description": "真实 YOLO/COCO data.yaml 路径", "key": "DATA_YAML", - "value": "", - "description": "YOLO data.yaml 路径", - "type": "string" + "type": "string", + "value": "" + }, + { + "description": "Ultralytics 模型名或权重路径", + "key": "MODEL_SIZE", + "type": "string", + "value": "yolo11n.pt" + }, + { + "description": "训练轮数", + "key": "EPOCHS", + "type": "number", + "value": "10" + }, + { + "description": "输入图片尺寸", + "key": "IMAGE_SIZE", + "type": "number", + "value": "640" + }, + { + "description": "训练 batch size", + "key": "BATCH_SIZE", + "type": "number", + "value": "8" + }, + { + "description": "后续检测推荐置信度", + "key": "CONF", + "type": "number", + "value": "0.25" + }, + { + "description": "关注类别,逗号分隔", + "key": "TARGET_LABELS", + "type": "string", + "value": "" }, { - "key": "RECOMMENDED_GPU", - "value": "1x NVIDIA A800 80GB", "description": "推荐 GPU 配置", - "type": "string" + "key": "RECOMMENDED_GPU", + "type": "string", + "value": "1x NVIDIA A800 80GB" + } + ], + "id": "operator-builtin-yolo-embodied-detection-train", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-yolo-embodied-detection-train:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-yolo-embodied-detection-train@sha256:52d3ddf838e032dc569a35cf1c937bd34241327a33e6f62efe91fadd1ea9a71a", + "inputs": [ + { + "description": "训练数据 manifest、切分清单或上游数据 artifact", + "id": "in-1", + "name": "dataset", + "transmission": "bus", + "type": "json" + } + ], + "name": "YOLO 具身目标检测训练", + "outputs": [ + { + "description": "训练得到的模型、权重、指标和日志 artifact", + "id": "out-1", + "name": "model", + "transmission": "bus", + "type": "file" } ], "resources": { "cpuLimit": "16", - "memoryLimit": "64Gi", "gpuLimit": "1", - "gpuType": "NVIDIA A800 80GB" + "gpuType": "NVIDIA A800 80GB", + "memoryLimit": "64Gi" }, + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": true, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_yolo_embodied_detection_train", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-yolo-embodied-detection-train", + "operatorSlug": "yolo_embodied_detection_train", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 1, + "recommendedGpuType": "NVIDIA A800 80GB", + "requiresGpu": true + }, + "runtimeDependencies": [ + "real-yolo-training-command", + "yolo-or-coco-dataset", + "gpu-a800" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_yolo_embodied_detection_train", "runtimeSource": "backend/operators/builtin/yolo_embodied_detection_train/operator.py", - "runtimeHandler": "builtin_yolo_embodied_detection_train" + "skillDefinition": "YOLO 具身目标检测训练算子;使用 Ultralytics API 读取真实 YOLO/COCO data.yaml,输出 best/last checkpoint 与 metrics.json。A800 对 YOLOv8/YOLOv10/YOLO11 中大模型训练充足,默认请求 1 卡。", + "type": "builtin" }, { - "id": "operator-builtin-yolo-video-target-clip", - "name": "YOLO 视频目标检测裁剪", - "type": "builtin", "category": "VIDEO", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "description": "使用真实 YOLO/ONNX 模型对视频执行目标检测,并按目标出现时间段裁剪片段。", - "inputs": [ - { - "id": "in-1", - "name": "videos", - "type": "json", - "description": "视频清单或切片清单", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "targetClips", - "type": "json", - "description": "目标出现片段清单和裁剪视频", - "transmission": "bus" - } - ], - "skillDefinition": "具体视频目标检测裁剪算子;基于已有 OpenCV DNN 路径执行真实 ONNX 推理,必须配置 DNN_MODEL_PATH,不再用笼统 motion/HOG 表达深度检测。A800 可用于批量视频检测加速。", "hyperparams": [ { - "key": "DNN_MODEL_PATH", - "value": "", "description": "真实 YOLO ONNX 模型路径", - "type": "string" + "key": "DNN_MODEL_PATH", + "type": "string", + "value": "" }, { - "key": "DNN_CLASS_NAMES", - "value": "", "description": "类别名,逗号分隔或使用 DNN_CLASS_NAMES_PATH", - "type": "string" + "key": "DNN_CLASS_NAMES", + "type": "string", + "value": "" }, { - "key": "TARGET_LABELS", - "value": "person,object", "description": "目标类别", - "type": "string" + "key": "TARGET_LABELS", + "type": "string", + "value": "person,object" }, { - "key": "FRAME_INTERVAL_SECONDS", - "value": "0.2", "description": "采样帧间隔", - "type": "string" + "key": "FRAME_INTERVAL_SECONDS", + "type": "string", + "value": "0.2" }, { - "key": "MIN_CONFIDENCE", - "value": "0.25", "description": "检测置信度阈值", - "type": "string" + "key": "MIN_CONFIDENCE", + "type": "string", + "value": "0.25" }, { - "key": "MAX_CLIPS", - "value": "50", "description": "最多输出片段数", - "type": "string" + "key": "MAX_CLIPS", + "type": "string", + "value": "50" + } + ], + "id": "operator-builtin-yolo-video-target-clip", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-yolo-video-target-clip:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-yolo-video-target-clip@sha256:36206dcc0ad5f80205273a0e983a66c4c47b65f2cecc94dba64d060016dafa67", + "inputs": [ + { + "description": "视频清单或切片清单", + "id": "in-1", + "name": "videos", + "transmission": "bus", + "type": "json" + } + ], + "name": "YOLO 视频目标检测裁剪", + "outputs": [ + { + "description": "目标出现片段清单和裁剪视频", + "id": "out-1", + "name": "targetClips", + "transmission": "bus", + "type": "json" } ], "resources": { "cpuLimit": "12", - "memoryLimit": "32Gi", "gpuLimit": "1", - "gpuType": "NVIDIA A800 80GB" + "gpuType": "NVIDIA A800 80GB", + "memoryLimit": "32Gi" }, + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": true, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_yolo_video_target_clip", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-yolo-video-target-clip", + "operatorSlug": "yolo_video_target_clip", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 1, + "recommendedGpuType": "NVIDIA A800 80GB", + "requiresGpu": true + }, + "runtimeDependencies": [ + "yolo-onnx-model", + "opencv-python-headless", + "numpy", + "ffmpeg", + "video-files", + "gpu-a800" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_yolo_video_target_clip", "runtimeSource": "backend/operators/builtin/yolo_video_target_clip/operator.py", - "runtimeHandler": "builtin_yolo_video_target_clip" + "skillDefinition": "具体视频目标检测裁剪算子;基于已有 OpenCV DNN 路径执行真实 ONNX 推理,必须配置 DNN_MODEL_PATH,不再用笼统 motion/HOG 表达深度检测。A800 可用于批量视频检测加速。", + "type": "builtin" }, { - "id": "operator-builtin-sim3d-asset-generate", - "name": "3D 资产仿真生成", - "type": "builtin", "category": "SIMULATION", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "description": "根据上游模型和样本统计生成可检查的 OBJ/MTL/glTF 3D 仿真资产。", + "hyperparams": [ + { + "description": "可选资产宽度;为空时按样本统计推导", + "key": "WIDTH_METERS", + "type": "number", + "value": "" + }, + { + "description": "可选资产深度;为空时按动作维度推导", + "key": "DEPTH_METERS", + "type": "number", + "value": "" + }, + { + "description": "可选资产高度;为空时按样本统计推导", + "key": "HEIGHT_METERS", + "type": "number", + "value": "" + } + ], + "id": "operator-builtin-sim3d-asset-generate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-sim3d-asset-generate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-sim3d-asset-generate@sha256:c56c68886041a19492d9427211dc4741574a95eacc3242d4f34c1adb6d2bcc36", "inputs": [ { + "description": "VLA 模型、样本清单或数据统计 artifact", "id": "in-1", "name": "model-or-samples", - "type": "json", - "description": "VLA 模型、样本清单或数据统计 artifact", - "transmission": "bus" + "transmission": "bus", + "type": "json" } ], + "name": "3D 资产仿真生成", "outputs": [ { + "description": "3D 资产 manifest、OBJ、MTL 和 glTF 文件", "id": "out-1", "name": "asset", - "type": "file", - "description": "3D 资产 manifest、OBJ、MTL 和 glTF 文件", - "transmission": "bus" + "transmission": "bus", + "type": "file" } ], - "skillDefinition": "从真实上游数据统计生成结构化 3D 资产文件,输出可追踪 manifest。", - "hyperparams": [ - { - "key": "WIDTH_METERS", - "value": "", - "description": "可选资产宽度;为空时按样本统计推导", - "type": "number" + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true }, - { - "key": "DEPTH_METERS", - "value": "", - "description": "可选资产深度;为空时按动作维度推导", - "type": "number" + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true }, - { - "key": "HEIGHT_METERS", - "value": "", - "description": "可选资产高度;为空时按样本统计推导", - "type": "number" - } - ], + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_sim3d_asset_generate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-sim3d-asset-generate", + "operatorSlug": "sim3d_asset_generate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "json-artifacts", + "obj-mtl-gltf-files" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_sim3d_asset_generate", "runtimeSource": "backend/operators/builtin/sim3d_asset_generate/operator.py", - "runtimeHandler": "builtin_sim3d_asset_generate" + "skillDefinition": "从真实上游数据统计生成结构化 3D 资产文件,输出可追踪 manifest。", + "type": "builtin" }, { - "id": "operator-builtin-vla-model-inference", - "name": "轻量线性策略推理", - "type": "builtin", "category": "AI", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "description": "加载轻量线性策略模型,对真实样本生成动作预测。", - "inputs": [ - { - "id": "in-1", - "name": "model-and-samples", - "type": "json", - "description": "VLA 策略模型、样本清单和可选 3D 资产 manifest", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "predictions", - "type": "json", - "description": "VLA 动作预测结果", - "transmission": "bus" - } - ], - "skillDefinition": "消费轻量线性策略模型和真实样本 artifact,输出动作向量预测;需要大模型推理时使用 OpenVLA 推理算子。", "hyperparams": [ { - "key": "MODEL_JSON", - "value": "", "description": "可选内联 VLA 模型 JSON;通常来自上游训练算子", - "type": "json" + "key": "MODEL_JSON", + "type": "json", + "value": "" }, { - "key": "INPUT_JSON", - "value": "", "description": "可选内联推理样本;通常来自上游数据算子", + "key": "INPUT_JSON", + "type": "json", + "value": "" + } + ], + "id": "operator-builtin-vla-model-inference", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-vla-model-inference:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-vla-model-inference@sha256:6e63bbeca4792ae37589211f3244c0545cbe2217e6f6bcf709b0575275d7d209", + "inputs": [ + { + "description": "VLA 策略模型、样本清单和可选 3D 资产 manifest", + "id": "in-1", + "name": "model-and-samples", + "transmission": "bus", + "type": "json" + } + ], + "name": "轻量线性策略推理", + "outputs": [ + { + "description": "VLA 动作预测结果", + "id": "out-1", + "name": "predictions", + "transmission": "bus", "type": "json" } ], - "runtimeSource": "backend/operators/builtin/vla_model_inference/operator.py", - "runtimeHandler": "builtin_vla_model_inference", "resources": { "cpuLimit": "2", "memoryLimit": "4Gi" - } + }, + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_vla_model_inference", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-vla-model-inference", + "operatorSlug": "vla_model_inference", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "vla-policy-model-json", + "vla-sample-action-manifest" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_vla_model_inference", + "runtimeSource": "backend/operators/builtin/vla_model_inference/operator.py", + "skillDefinition": "消费轻量线性策略模型和真实样本 artifact,输出动作向量预测;需要大模型推理时使用 OpenVLA 推理算子。", + "type": "builtin" }, { - "id": "operator-builtin-vla-model-evaluate", - "name": "轻量策略评测", - "type": "builtin", "category": "AI", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "description": "对轻量策略推理结果与真实 action 进行配对评测,输出 RMSE/MAE 等指标。", - "inputs": [ - { - "id": "in-1", - "name": "predictions-and-labels", - "type": "json", - "description": "VLA 推理结果和带真实 action 的样本清单", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "report", - "type": "json", - "description": "VLA 模型评测报告", - "transmission": "bus" - } - ], - "skillDefinition": "读取真实预测和真实 action,计算 RMSE、MAE、最大绝对误差并给出验收状态。", "hyperparams": [ { - "key": "MIN_EVAL_SAMPLES", - "value": "1", "description": "最少可评测样本数", - "type": "number" + "key": "MIN_EVAL_SAMPLES", + "type": "number", + "value": "1" }, { - "key": "MAX_RMSE", - "value": "10.0", "description": "评测通过的最大 RMSE", - "type": "number" + "key": "MAX_RMSE", + "type": "number", + "value": "10.0" + } + ], + "id": "operator-builtin-vla-model-evaluate", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-vla-model-evaluate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-vla-model-evaluate@sha256:d940646dabe0e5b7b950e1f32a0ce36110d9dbc379bd123dbb761b8f2e896464", + "inputs": [ + { + "description": "VLA 推理结果和带真实 action 的样本清单", + "id": "in-1", + "name": "predictions-and-labels", + "transmission": "bus", + "type": "json" + } + ], + "name": "轻量策略评测", + "outputs": [ + { + "description": "VLA 模型评测报告", + "id": "out-1", + "name": "report", + "transmission": "bus", + "type": "json" } ], - "runtimeSource": "backend/operators/builtin/vla_model_evaluate/operator.py", - "runtimeHandler": "builtin_vla_model_evaluate", "resources": { "cpuLimit": "1", "memoryLimit": "2Gi" - } + }, + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_vla_model_evaluate", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-vla-model-evaluate", + "operatorSlug": "vla_model_evaluate", + "parameters": { + "commandWrapperAllowed": false, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "vla-inference-result-json", + "vla-sample-action-manifest" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_vla_model_evaluate", + "runtimeSource": "backend/operators/builtin/vla_model_evaluate/operator.py", + "skillDefinition": "读取真实预测和真实 action,计算 RMSE、MAE、最大绝对误差并给出验收状态。", + "type": "builtin" }, { - "id": "operator-builtin-shell-command", - "name": "Shell 命令执行", - "type": "builtin", "category": "UTIL", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], "description": "在 operator-runner 容器中真实执行 shell 命令并收集输出文件。", - "inputs": [ - { - "id": "in-1", - "name": "input", - "type": "json", - "description": "输入数据", - "transmission": "bus" - } - ], - "outputs": [ - { - "id": "out-1", - "name": "result", - "type": "file", - "description": "执行结果文件", - "transmission": "bus" - } - ], - "skillDefinition": "执行 COMMAND 或 SCRIPT,自动收集输出目录产物并生成预览元数据。", "hyperparams": [ { - "key": "COMMAND", - "value": "", "description": "容器内执行的命令", - "type": "string" + "key": "COMMAND", + "type": "string", + "value": "" }, { - "key": "SCRIPT", - "value": "", "description": "可执行脚本内容", - "type": "string" + "key": "SCRIPT", + "type": "string", + "value": "" } ], + "id": "operator-builtin-shell-command", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-shell-command:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-shell-command@sha256:47f7039f4ddcf58fb51cf38f1e4fe8fd7c96e8bcad2ea745c88ee0d9e1423f51", + "inputs": [ + { + "description": "输入数据", + "id": "in-1", + "name": "input", + "transmission": "bus", + "type": "json" + } + ], + "name": "Shell 命令执行", + "outputs": [ + { + "description": "执行结果文件", + "id": "out-1", + "name": "result", + "transmission": "bus", + "type": "file" + } + ], + "runtimeContract": { + "acceptance": { + "requiresArtifactChecksum": true, + "requiresFixtureChecksum": true, + "requiresGpuEvidence": false, + "requiresRealFixture": true, + "requiresWorkflowTransfer": true + }, + "artifacts": { + "requiresChecksum": true, + "requiresLineage": true, + "requiresManifest": true + }, + "entrypoint": "/app/backend/operator-runner/main.py", + "handler": "builtin_shell_command", + "imagePolicy": { + "allowsSharedBaseLayer": true, + "forbidsSharedOperatorRunnerFinalImage": true, + "independentFinalImage": true + }, + "operatorId": "operator-builtin-shell-command", + "operatorSlug": "shell_command", + "parameters": { + "commandWrapperAllowed": true, + "source": "hyperparams", + "structured": true + }, + "resources": { + "minGpuCount": 0, + "recommendedGpuType": "", + "requiresGpu": false + }, + "runtimeDependencies": [ + "container-shell" + ], + "schemaVersion": "droboticflow.operator.runtime/v1" + }, + "runtimeHandler": "builtin_shell_command", "runtimeSource": "backend/operators/builtin/shell_command/operator.py", - "runtimeHandler": "builtin_shell_command" + "skillDefinition": "执行 COMMAND 或 SCRIPT,自动收集输出目录产物并生成预览元数据。", + "type": "builtin" } ] diff --git a/internal/api/operator_image_archive.go b/internal/api/operator_image_archive.go index 65527cfa..adea0563 100644 --- a/internal/api/operator_image_archive.go +++ b/internal/api/operator_image_archive.go @@ -238,7 +238,7 @@ func (server *server) handleOperatorImageArchiveImport(writer http.ResponseWrite return } payload.Version = normalizeOperatorVersion(payload.Version, tag) - if issues := validateOperatorMetadata(payload.Name, payload.Type, payload.Category, payload.Description, payload.Version, payload.Inputs, payload.Outputs, payload.Resources); len(issues) > 0 { + if issues := validateOperatorMetadata(payload.Name, payload.Type, payload.Category, payload.Description, payload.Version, payload.Inputs, payload.Outputs, payload.Resources, payload.ContainerCommand, payload.ContainerArgs); len(issues) > 0 { renderValidationError(writer, issues...) return } diff --git a/internal/api/poc_operator_workflow_acceptance.go b/internal/api/poc_operator_workflow_acceptance.go index 8004ec60..578c3999 100644 --- a/internal/api/poc_operator_workflow_acceptance.go +++ b/internal/api/poc_operator_workflow_acceptance.go @@ -23,20 +23,43 @@ type pocOperatorWorkflowAcceptanceRequest struct { } type pocOperatorWorkflowAcceptanceRun struct { - Batch string `json:"batch"` - WorkflowID string `json:"workflowId"` - Workflow *domain.WorkflowRecord `json:"workflow"` - RunID string `json:"runId"` - Run domain.RunRecord `json:"run"` - OperatorIDs []string `json:"operatorIds"` + Batch string `json:"batch"` + WorkflowID string `json:"workflowId"` + Workflow *domain.WorkflowRecord `json:"workflow"` + RunID string `json:"runId"` + Run domain.RunRecord `json:"run"` + OperatorIDs []string `json:"operatorIds"` + ImageEvidence []pocOperatorWorkflowImageEvidence `json:"imageEvidence"` +} + +type pocOperatorWorkflowImageEvidence struct { + OperatorID string `json:"operatorId"` + Image string `json:"image"` + ImageID string `json:"imageId"` + ContainerCommand []string `json:"containerCommand"` + ContainerArgs []string `json:"containerArgs"` + RequiresGPU bool `json:"requiresGpu"` +} + +type pocOperatorWorkflowAcceptanceSummary struct { + ContractVersion string `json:"contractVersion"` + OperatorCount int `json:"operatorCount"` + IndependentImageCoverage int `json:"independentImageCoverage"` + IndependentImageRequired int `json:"independentImageRequired"` + CommandWrapperOnlyOperatorCount int `json:"commandWrapperOnlyOperatorCount"` + GeneratedFixtureCount int `json:"generatedFixtureCount"` + MockFixtureCount int `json:"mockFixtureCount"` + GPUOperatorCount int `json:"gpuOperatorCount"` + StrictReady bool `json:"strictReady"` } type pocOperatorWorkflowAcceptanceResponse struct { - Project *domain.ProjectRecord `json:"project"` - FixtureManifestPath string `json:"fixtureManifestPath"` - OperatorCount int `json:"operatorCount"` - Runs []pocOperatorWorkflowAcceptanceRun `json:"runs"` - Batches map[string][]string `json:"batches"` + Project *domain.ProjectRecord `json:"project"` + FixtureManifestPath string `json:"fixtureManifestPath"` + OperatorCount int `json:"operatorCount"` + Summary pocOperatorWorkflowAcceptanceSummary `json:"summary"` + Runs []pocOperatorWorkflowAcceptanceRun `json:"runs"` + Batches map[string][]string `json:"batches"` } type pocOperatorAcceptanceFixturePaths struct { @@ -224,9 +247,24 @@ func configurePocOperatorHyperparams(operator domain.OperatorRecord, fixtures po case "builtin_model_inference": items = setHyperparam(items, "MODEL_PATH", fixtures.Model) items = setHyperparam(items, "INPUT_JSON", fmt.Sprintf(`{"items":[{"id":"acceptance-a","score":0.95,"quality":{"clarity":0.95,"shake":0.01}},{"id":"acceptance-b","score":0.42,"quality":{"clarity":0.42,"shake":0.2}}]}`)) - case "builtin_act_policy_train", "builtin_diffusion_policy_train", "builtin_openvla_finetune", "builtin_openvla_inference", "builtin_yolo_embodied_detection_train": - setSource(fixtures.JSON) - items = setHyperparam(items, "POC_ACCEPTANCE_MODE", "true") + case "builtin_act_policy_train", "builtin_diffusion_policy_train": + items = setHyperparam(items, "DATASET_PATH", fixtures.JSON) + items = setHyperparam(items, "STATE_FIELD", "score") + items = setHyperparam(items, "ACTION_FIELD", "score") + items = setHyperparam(items, "EPOCHS", "1") + items = setHyperparam(items, "BATCH_SIZE", "2") + case "builtin_openvla_finetune": + items = setHyperparam(items, "DATASET_PATH", fixtures.JSON) + items = setHyperparam(items, "TEXT_FIELD", "label") + items = setHyperparam(items, "TARGET_FIELD", "id") + items = setHyperparam(items, "EPOCHS", "1") + items = setHyperparam(items, "BATCH_SIZE", "1") + case "builtin_openvla_inference": + items = setHyperparam(items, "PROMPT", "pick the visible target") + case "builtin_yolo_embodied_detection_train": + if value := strings.TrimSpace(fixtures.Extra["yolo"]); value != "" { + items = setHyperparam(items, "DATA_YAML", value) + } case "builtin_json_extract": items = setHyperparam(items, "FIELD_PATH", "items.0.label") case "builtin_json_filter", "builtin_condition_router": @@ -558,20 +596,23 @@ func workflowNodeFromOperator(operator domain.OperatorRecord, fixtures pocOperat ID: operatorNodeID(operator), Type: "operator", Data: domain.WorkflowNodeData{ - Label: operator.Name, - Category: operator.Category, - Description: operator.Description, - Inputs: clonePortDefinitions(operator.Inputs), - Outputs: clonePortDefinitions(operator.Outputs), - OperatorID: operator.ID, - OperatorVersion: operator.Version, - Image: operator.Image, - Hyperparams: hyperparams, - ConnectionID: connectionID, - Bucket: bucket, - CPU: operatorResourceCPULimit(operator.Resources), - Memory: operatorResourceMemoryLimit(operator.Resources), - GPU: operatorResourceGPULimit(operator.Resources), + Label: operator.Name, + Category: operator.Category, + Description: operator.Description, + Inputs: clonePortDefinitions(operator.Inputs), + Outputs: clonePortDefinitions(operator.Outputs), + OperatorID: operator.ID, + OperatorVersion: operator.Version, + OperatorImageID: operator.ImageID, + Image: operator.Image, + ContainerCommand: cloneStringSlice(operator.ContainerCommand), + ContainerArgs: cloneStringSlice(operator.ContainerArgs), + Hyperparams: hyperparams, + ConnectionID: connectionID, + Bucket: bucket, + CPU: operatorResourceCPULimit(operator.Resources), + Memory: operatorResourceMemoryLimit(operator.Resources), + GPU: operatorResourceGPULimit(operator.Resources), }, Position: domain.Position{X: x, Y: y}, } @@ -598,6 +639,74 @@ func operatorResourceGPULimit(resources *domain.OperatorResources) string { return strings.TrimSpace(resources.GPULimit) } +func pocOperatorHasCommandWrapper(operator domain.OperatorRecord) bool { + if operator.ID == "operator-builtin-shell-command" { + return false + } + for _, param := range operator.Hyperparams { + key := strings.ToUpper(strings.TrimSpace(param.Key)) + if key == "COMMAND" || key == "SCRIPT" { + return true + } + } + return false +} + +func pocOperatorRequiresGPU(operator domain.OperatorRecord) bool { + return operator.Resources != nil && strings.TrimSpace(operator.Resources.GPULimit) != "" +} + +func pocOperatorHasIndependentImage(operator domain.OperatorRecord) bool { + image := strings.TrimSpace(operator.Image) + imageID := strings.TrimSpace(operator.ImageID) + if image == "" || strings.Contains(image, "droboticflow-operator-runner") || !strings.Contains(image, "droboticflow-operator-") { + return false + } + if !strings.Contains(imageID, "@sha256:") { + return false + } + if len(operator.ContainerCommand) == 0 { + return false + } + return true +} + +func buildPocOperatorWorkflowImageEvidence(operator domain.OperatorRecord) pocOperatorWorkflowImageEvidence { + return pocOperatorWorkflowImageEvidence{ + OperatorID: operator.ID, + Image: strings.TrimSpace(operator.Image), + ImageID: strings.TrimSpace(operator.ImageID), + ContainerCommand: cloneStringSlice(operator.ContainerCommand), + ContainerArgs: cloneStringSlice(operator.ContainerArgs), + RequiresGPU: pocOperatorRequiresGPU(operator), + } +} + +func buildPocOperatorWorkflowAcceptanceSummary(operators []domain.OperatorRecord) pocOperatorWorkflowAcceptanceSummary { + summary := pocOperatorWorkflowAcceptanceSummary{ + ContractVersion: "poc-operator-workflows/v2", + OperatorCount: len(operators), + IndependentImageRequired: len(operators), + } + for _, operator := range operators { + if pocOperatorHasIndependentImage(operator) { + summary.IndependentImageCoverage++ + } + if pocOperatorHasCommandWrapper(operator) { + summary.CommandWrapperOnlyOperatorCount++ + } + if pocOperatorRequiresGPU(operator) { + summary.GPUOperatorCount++ + } + } + summary.StrictReady = summary.OperatorCount > 0 && + summary.IndependentImageCoverage == summary.IndependentImageRequired && + summary.CommandWrapperOnlyOperatorCount == 0 && + summary.GeneratedFixtureCount == 0 && + summary.MockFixtureCount == 0 + return summary +} + func clonePortDefinitions(items []domain.PortDefinition) []domain.PortDefinition { if len(items) == 0 { return nil @@ -609,6 +718,15 @@ func clonePortDefinitions(items []domain.PortDefinition) []domain.PortDefinition return cloned } +func cloneStringSlice(items []string) []string { + if len(items) == 0 { + return nil + } + cloned := make([]string, len(items)) + copy(cloned, items) + return cloned +} + func groupPocOperatorAcceptanceBatches(operators []domain.OperatorRecord) map[string][]domain.OperatorRecord { batches := map[string][]domain.OperatorRecord{ "data-io-core": {}, @@ -878,6 +996,7 @@ func (server *server) handlePocOperatorWorkflowAcceptance(writer http.ResponseWr Project: sanitizeProject(project), FixtureManifestPath: manifestPath, OperatorCount: len(operators), + Summary: buildPocOperatorWorkflowAcceptanceSummary(operators), Runs: []pocOperatorWorkflowAcceptanceRun{}, Batches: map[string][]string{}, } @@ -898,17 +1017,20 @@ func (server *server) handlePocOperatorWorkflowAcceptance(writer http.ResponseWr return } operatorIDs := make([]string, 0, len(batchOperators)) + imageEvidence := make([]pocOperatorWorkflowImageEvidence, 0, len(batchOperators)) for _, operator := range batchOperators { operatorIDs = append(operatorIDs, operator.ID) + imageEvidence = append(imageEvidence, buildPocOperatorWorkflowImageEvidence(operator)) } response.Batches[batchName] = operatorIDs response.Runs = append(response.Runs, pocOperatorWorkflowAcceptanceRun{ - Batch: batchName, - WorkflowID: workflow.ID, - Workflow: sanitizeWorkflow(workflow), - RunID: run.ID, - Run: runForResponse(request, *run), - OperatorIDs: operatorIDs, + Batch: batchName, + WorkflowID: workflow.ID, + Workflow: sanitizeWorkflow(workflow), + RunID: run.ID, + Run: runForResponse(request, *run), + OperatorIDs: operatorIDs, + ImageEvidence: imageEvidence, }) } renderJSON(writer, http.StatusAccepted, response) diff --git a/internal/api/poc_operator_workflow_acceptance_test.go b/internal/api/poc_operator_workflow_acceptance_test.go index 3dda13d8..ede23196 100644 --- a/internal/api/poc_operator_workflow_acceptance_test.go +++ b/internal/api/poc_operator_workflow_acceptance_test.go @@ -1,6 +1,7 @@ package api import ( + "strings" "testing" "droboticflow/internal/domain" @@ -290,6 +291,13 @@ func TestWorkflowNodeFromOperatorCopiesDefaultGPUResources(t *testing.T) { Name: "ACT 策略训练", Type: "builtin", Category: "AI", + Image: "ccr.example/droboticflow-operator-act-policy-train:v1", + ImageID: "ccr.example/droboticflow-operator-act-policy-train@sha256:abc123", + ContainerCommand: []string{ + "python", + "/app/backend/operator-runner/main.py", + }, + ContainerArgs: []string{"--strict"}, Resources: &domain.OperatorResources{ CPULimit: "16", MemoryLimit: "64Gi", @@ -300,8 +308,23 @@ func TestWorkflowNodeFromOperatorCopiesDefaultGPUResources(t *testing.T) { if node.Data.CPU != "16" || node.Data.Memory != "64Gi" || node.Data.GPU != "1" { t.Fatalf("node resources = cpu %q memory %q gpu %q", node.Data.CPU, node.Data.Memory, node.Data.GPU) } - if mode := hyperparamValue(node.Data.Hyperparams, "POC_ACCEPTANCE_MODE"); mode != "true" { - t.Fatalf("POC_ACCEPTANCE_MODE = %q", mode) + if node.Data.Image != "ccr.example/droboticflow-operator-act-policy-train:v1" { + t.Fatalf("node image = %q", node.Data.Image) + } + if node.Data.OperatorImageID != "ccr.example/droboticflow-operator-act-policy-train@sha256:abc123" { + t.Fatalf("node image id = %q", node.Data.OperatorImageID) + } + if got := strings.Join(node.Data.ContainerCommand, " "); got != "python /app/backend/operator-runner/main.py" { + t.Fatalf("container command = %q", got) + } + if got := strings.Join(node.Data.ContainerArgs, " "); got != "--strict" { + t.Fatalf("container args = %q", got) + } + if mode := hyperparamValue(node.Data.Hyperparams, "POC_ACCEPTANCE_MODE"); mode != "" { + t.Fatalf("POC_ACCEPTANCE_MODE must not be used for production GPU operators, got %q", mode) + } + if datasetPath := hyperparamValue(node.Data.Hyperparams, "DATASET_PATH"); datasetPath == "" { + t.Fatalf("DATASET_PATH should be configured for structured GPU acceptance") } } diff --git a/internal/api/server.go b/internal/api/server.go index 8d42bd66..998090d7 100644 --- a/internal/api/server.go +++ b/internal/api/server.go @@ -3669,6 +3669,41 @@ func normalizeOperatorResources(resources *domain.OperatorResources) *domain.Ope } } +func normalizeContainerCommand(items []string) []string { + if len(items) == 0 { + return nil + } + normalized := make([]string, 0, len(items)) + for _, item := range items { + value := strings.TrimSpace(item) + if value != "" { + normalized = append(normalized, value) + } + } + if len(normalized) == 0 { + return nil + } + return normalized +} + +func validateContainerCommand(field string, items []string) []validationIssue { + issues := make([]validationIssue, 0) + for index, item := range items { + prefix := field + "[" + strconv.Itoa(index) + "]" + if strings.TrimSpace(item) == "" { + issues = append(issues, validationIssue{Field: prefix, Message: "value is required"}) + continue + } + if len(item) > 512 { + issues = append(issues, validationIssue{Field: prefix, Message: "value must be at most 512 characters"}) + } + if strings.ContainsAny(item, "\x00\r\n") { + issues = append(issues, validationIssue{Field: prefix, Message: "value must not contain control characters"}) + } + } + return issues +} + func validateOperatorResources(resources *domain.OperatorResources) []validationIssue { if resources == nil { return nil @@ -3695,7 +3730,7 @@ func validateOperatorResources(resources *domain.OperatorResources) []validation return nil } -func validateOperatorMetadata(name string, operatorType string, category string, description string, version string, inputs []domain.PortDefinition, outputs []domain.PortDefinition, resources *domain.OperatorResources) []validationIssue { +func validateOperatorMetadata(name string, operatorType string, category string, description string, version string, inputs []domain.PortDefinition, outputs []domain.PortDefinition, resources *domain.OperatorResources, containerCommand []string, containerArgs []string) []validationIssue { issues := make([]validationIssue, 0) if name == "" { issues = append(issues, validationIssue{Field: "name", Message: "name is required"}) @@ -3713,6 +3748,8 @@ func validateOperatorMetadata(name string, operatorType string, category string, issues = append(issues, validatePortDefinitions("inputs", inputs)...) issues = append(issues, validatePortDefinitions("outputs", outputs)...) issues = append(issues, validateOperatorResources(resources)...) + issues = append(issues, validateContainerCommand("containerCommand", containerCommand)...) + issues = append(issues, validateContainerCommand("containerArgs", containerArgs)...) return issues } @@ -3733,12 +3770,14 @@ func (server *server) handleOperatorsCreate(writer http.ResponseWriter, request payload.Version = normalizeOperatorVersion(payload.Version, "v1") payload.Image = strings.TrimSpace(payload.Image) payload.ImageID = strings.TrimSpace(payload.ImageID) + payload.ContainerCommand = normalizeContainerCommand(payload.ContainerCommand) + payload.ContainerArgs = normalizeContainerCommand(payload.ContainerArgs) payload.SkillDefinition = strings.TrimSpace(payload.SkillDefinition) payload.Resources = normalizeOperatorResources(payload.Resources) if payload.ImageID == "" { payload.ImageID = payload.Image } - if issues := validateOperatorMetadata(payload.Name, payload.Type, payload.Category, payload.Description, payload.Version, payload.Inputs, payload.Outputs, payload.Resources); len(issues) > 0 { + if issues := validateOperatorMetadata(payload.Name, payload.Type, payload.Category, payload.Description, payload.Version, payload.Inputs, payload.Outputs, payload.Resources, payload.ContainerCommand, payload.ContainerArgs); len(issues) > 0 { renderValidationError(writer, issues...) return } @@ -3772,19 +3811,22 @@ func (server *server) handleOperatorsGet(writer http.ResponseWriter, request *ht } type operatorPatch struct { - Name *string `json:"name"` - Type *string `json:"type"` - Category *string `json:"category"` - Description *string `json:"description"` - Inputs *[]domain.PortDefinition `json:"inputs"` - Outputs *[]domain.PortDefinition `json:"outputs"` - ConfigSchema *map[string]any `json:"configSchema"` - Version *string `json:"version"` - Image *string `json:"image"` - ImageID *string `json:"imageId"` - SkillDefinition *string `json:"skillDefinition"` - Hyperparams *[]domain.HyperParam `json:"hyperparams"` - Resources *domain.OperatorResources `json:"resources"` + Name *string `json:"name"` + Type *string `json:"type"` + Category *string `json:"category"` + Description *string `json:"description"` + Inputs *[]domain.PortDefinition `json:"inputs"` + Outputs *[]domain.PortDefinition `json:"outputs"` + ConfigSchema *map[string]any `json:"configSchema"` + Version *string `json:"version"` + Image *string `json:"image"` + ImageID *string `json:"imageId"` + ContainerCommand *[]string `json:"containerCommand"` + ContainerArgs *[]string `json:"containerArgs"` + RuntimeContract *map[string]any `json:"runtimeContract"` + SkillDefinition *string `json:"skillDefinition"` + Hyperparams *[]domain.HyperParam `json:"hyperparams"` + Resources *domain.OperatorResources `json:"resources"` } func (server *server) handleOperatorsUpdate(writer http.ResponseWriter, request *http.Request) { @@ -3846,6 +3888,15 @@ func (server *server) handleOperatorsUpdate(writer http.ResponseWriter, request } else if payload.Image != nil && next.ImageID == "" { next.ImageID = next.Image } + if payload.ContainerCommand != nil { + next.ContainerCommand = normalizeContainerCommand(*payload.ContainerCommand) + } + if payload.ContainerArgs != nil { + next.ContainerArgs = normalizeContainerCommand(*payload.ContainerArgs) + } + if payload.RuntimeContract != nil { + next.RuntimeContract = *payload.RuntimeContract + } if payload.SkillDefinition != nil { next.SkillDefinition = strings.TrimSpace(*payload.SkillDefinition) } @@ -3861,7 +3912,7 @@ func (server *server) handleOperatorsUpdate(writer http.ResponseWriter, request next.ImageID = next.Image } } - if issues := validateOperatorMetadata(next.Name, next.Type, next.Category, next.Description, next.Version, next.Inputs, next.Outputs, next.Resources); len(issues) > 0 { + if issues := validateOperatorMetadata(next.Name, next.Type, next.Category, next.Description, next.Version, next.Inputs, next.Outputs, next.Resources, next.ContainerCommand, next.ContainerArgs); len(issues) > 0 { renderValidationError(writer, issues...) return } diff --git a/internal/domain/types.go b/internal/domain/types.go index c30f07ef..5e35b353 100644 --- a/internal/domain/types.go +++ b/internal/domain/types.go @@ -204,23 +204,26 @@ type OperatorResources struct { } type OperatorRecord struct { - ID string `json:"id" bson:"id"` - OrganizationID string `json:"organizationId,omitempty" bson:"organizationId,omitempty"` - Name string `json:"name" bson:"name"` - Type string `json:"type" bson:"type"` - Category string `json:"category" bson:"category"` - Description string `json:"description" bson:"description"` - Inputs []PortDefinition `json:"inputs" bson:"inputs"` - Outputs []PortDefinition `json:"outputs" bson:"outputs"` - ConfigSchema map[string]any `json:"configSchema,omitempty" bson:"configSchema,omitempty"` - Version string `json:"version,omitempty" bson:"version,omitempty"` - Image string `json:"image,omitempty" bson:"image,omitempty"` - ImageID string `json:"imageId,omitempty" bson:"imageId,omitempty"` - SkillDefinition string `json:"skillDefinition,omitempty" bson:"skillDefinition,omitempty"` - Hyperparams []HyperParam `json:"hyperparams,omitempty" bson:"hyperparams,omitempty"` - Resources *OperatorResources `json:"resources,omitempty" bson:"resources,omitempty"` - CreatedAt string `json:"createdAt" bson:"createdAt"` - UpdatedAt string `json:"updatedAt" bson:"updatedAt"` + ID string `json:"id" bson:"id"` + OrganizationID string `json:"organizationId,omitempty" bson:"organizationId,omitempty"` + Name string `json:"name" bson:"name"` + Type string `json:"type" bson:"type"` + Category string `json:"category" bson:"category"` + Description string `json:"description" bson:"description"` + Inputs []PortDefinition `json:"inputs" bson:"inputs"` + Outputs []PortDefinition `json:"outputs" bson:"outputs"` + ConfigSchema map[string]any `json:"configSchema,omitempty" bson:"configSchema,omitempty"` + Version string `json:"version,omitempty" bson:"version,omitempty"` + Image string `json:"image,omitempty" bson:"image,omitempty"` + ImageID string `json:"imageId,omitempty" bson:"imageId,omitempty"` + ContainerCommand []string `json:"containerCommand,omitempty" bson:"containerCommand,omitempty"` + ContainerArgs []string `json:"containerArgs,omitempty" bson:"containerArgs,omitempty"` + RuntimeContract map[string]any `json:"runtimeContract,omitempty" bson:"runtimeContract,omitempty"` + SkillDefinition string `json:"skillDefinition,omitempty" bson:"skillDefinition,omitempty"` + Hyperparams []HyperParam `json:"hyperparams,omitempty" bson:"hyperparams,omitempty"` + Resources *OperatorResources `json:"resources,omitempty" bson:"resources,omitempty"` + CreatedAt string `json:"createdAt" bson:"createdAt"` + UpdatedAt string `json:"updatedAt" bson:"updatedAt"` } type BuiltinOperatorDeletionRecord struct { @@ -802,20 +805,23 @@ type CreateTemplateInput struct { } type CreateOperatorInput struct { - OrganizationID string `json:"organizationId,omitempty"` - Name string `json:"name"` - Type string `json:"type"` - Category string `json:"category"` - Description string `json:"description"` - Inputs []PortDefinition `json:"inputs"` - Outputs []PortDefinition `json:"outputs"` - ConfigSchema map[string]any `json:"configSchema,omitempty"` - Version string `json:"version,omitempty"` - Image string `json:"image,omitempty"` - ImageID string `json:"imageId,omitempty"` - SkillDefinition string `json:"skillDefinition,omitempty"` - Hyperparams []HyperParam `json:"hyperparams,omitempty"` - Resources *OperatorResources `json:"resources,omitempty"` + OrganizationID string `json:"organizationId,omitempty"` + Name string `json:"name"` + Type string `json:"type"` + Category string `json:"category"` + Description string `json:"description"` + Inputs []PortDefinition `json:"inputs"` + Outputs []PortDefinition `json:"outputs"` + ConfigSchema map[string]any `json:"configSchema,omitempty"` + Version string `json:"version,omitempty"` + Image string `json:"image,omitempty"` + ImageID string `json:"imageId,omitempty"` + ContainerCommand []string `json:"containerCommand,omitempty"` + ContainerArgs []string `json:"containerArgs,omitempty"` + RuntimeContract map[string]any `json:"runtimeContract,omitempty"` + SkillDefinition string `json:"skillDefinition,omitempty"` + Hyperparams []HyperParam `json:"hyperparams,omitempty"` + Resources *OperatorResources `json:"resources,omitempty"` } type CreateUserInput struct { diff --git a/internal/store/memory.go b/internal/store/memory.go index 7e68a9fe..f7f1ce98 100644 --- a/internal/store/memory.go +++ b/internal/store/memory.go @@ -464,23 +464,26 @@ func (store *MemoryStore) CreateOperator(_ context.Context, input domain.CreateO defer store.mu.Unlock() now := nowISO() item := domain.OperatorRecord{ - ID: domain.NewID("operator"), - OrganizationID: input.OrganizationID, - Name: input.Name, - Type: input.Type, - Category: input.Category, - Description: input.Description, - Inputs: clone(input.Inputs), - Outputs: clone(input.Outputs), - ConfigSchema: clone(input.ConfigSchema), - Version: input.Version, - Image: input.Image, - ImageID: input.ImageID, - SkillDefinition: input.SkillDefinition, - Hyperparams: clone(input.Hyperparams), - Resources: clone(input.Resources), - CreatedAt: now, - UpdatedAt: now, + ID: domain.NewID("operator"), + OrganizationID: input.OrganizationID, + Name: input.Name, + Type: input.Type, + Category: input.Category, + Description: input.Description, + Inputs: clone(input.Inputs), + Outputs: clone(input.Outputs), + ConfigSchema: clone(input.ConfigSchema), + Version: input.Version, + Image: input.Image, + ImageID: input.ImageID, + ContainerCommand: clone(input.ContainerCommand), + ContainerArgs: clone(input.ContainerArgs), + RuntimeContract: clone(input.RuntimeContract), + SkillDefinition: input.SkillDefinition, + Hyperparams: clone(input.Hyperparams), + Resources: clone(input.Resources), + CreatedAt: now, + UpdatedAt: now, } store.operators[item.ID] = clone(item) cloned := clone(item) diff --git a/internal/store/mongo.go b/internal/store/mongo.go index f4b0d215..79cbfc1e 100644 --- a/internal/store/mongo.go +++ b/internal/store/mongo.go @@ -324,23 +324,26 @@ func (store *MongoStore) GetOperator(ctx context.Context, id string) (*domain.Op func (store *MongoStore) CreateOperator(ctx context.Context, input domain.CreateOperatorInput) (*domain.OperatorRecord, error) { now := nowISO() item := domain.OperatorRecord{ - ID: domain.NewID("operator"), - OrganizationID: input.OrganizationID, - Name: input.Name, - Type: input.Type, - Category: input.Category, - Description: input.Description, - Inputs: input.Inputs, - Outputs: input.Outputs, - ConfigSchema: input.ConfigSchema, - Version: input.Version, - Image: input.Image, - ImageID: input.ImageID, - SkillDefinition: input.SkillDefinition, - Hyperparams: input.Hyperparams, - Resources: input.Resources, - CreatedAt: now, - UpdatedAt: now, + ID: domain.NewID("operator"), + OrganizationID: input.OrganizationID, + Name: input.Name, + Type: input.Type, + Category: input.Category, + Description: input.Description, + Inputs: input.Inputs, + Outputs: input.Outputs, + ConfigSchema: input.ConfigSchema, + Version: input.Version, + Image: input.Image, + ImageID: input.ImageID, + ContainerCommand: input.ContainerCommand, + ContainerArgs: input.ContainerArgs, + RuntimeContract: input.RuntimeContract, + SkillDefinition: input.SkillDefinition, + Hyperparams: input.Hyperparams, + Resources: input.Resources, + CreatedAt: now, + UpdatedAt: now, } if err := store.UpsertOperator(ctx, item); err != nil { return nil, err diff --git a/operator-images/builtin/act_policy_train/Dockerfile b/operator-images/builtin/act_policy_train/Dockerfile new file mode 100644 index 00000000..79f2ebf6 --- /dev/null +++ b/operator-images/builtin/act_policy_train/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM pytorch/pytorch:2.5.1-cuda12.4-cudnn9-runtime + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-act-policy-train \ + DROBOTICFLOW_OPERATOR_SLUG=act_policy_train \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_act_policy_train \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py torch torchvision transformers accelerate safetensors +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/action_state_quality_validate/Dockerfile b/operator-images/builtin/action_state_quality_validate/Dockerfile new file mode 100644 index 00000000..615c0a46 --- /dev/null +++ b/operator-images/builtin/action_state_quality_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-action-state-quality-validate \ + DROBOTICFLOW_OPERATOR_SLUG=action_state_quality_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_action_state_quality_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/agibot_world2026_validate/Dockerfile b/operator-images/builtin/agibot_world2026_validate/Dockerfile new file mode 100644 index 00000000..5080200c --- /dev/null +++ b/operator-images/builtin/agibot_world2026_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-agibot-world2026-validate \ + DROBOTICFLOW_OPERATOR_SLUG=agibot_world2026_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_agibot_world2026_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/aloha_hdf5_validate/Dockerfile b/operator-images/builtin/aloha_hdf5_validate/Dockerfile new file mode 100644 index 00000000..51f4396d --- /dev/null +++ b/operator-images/builtin/aloha_hdf5_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-aloha-hdf5-validate \ + DROBOTICFLOW_OPERATOR_SLUG=aloha_hdf5_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_aloha_hdf5_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/audit_sample/Dockerfile b/operator-images/builtin/audit_sample/Dockerfile new file mode 100644 index 00000000..c719310c --- /dev/null +++ b/operator-images/builtin/audit_sample/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-audit-sample \ + DROBOTICFLOW_OPERATOR_SLUG=audit_sample \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_audit_sampling \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/bop_dataset_validate/Dockerfile b/operator-images/builtin/bop_dataset_validate/Dockerfile new file mode 100644 index 00000000..bccc90a2 --- /dev/null +++ b/operator-images/builtin/bop_dataset_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-bop-dataset-validate \ + DROBOTICFLOW_OPERATOR_SLUG=bop_dataset_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_bop_dataset_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/bos_incremental_scan/Dockerfile b/operator-images/builtin/bos_incremental_scan/Dockerfile new file mode 100644 index 00000000..3eb1931a --- /dev/null +++ b/operator-images/builtin/bos_incremental_scan/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-bos-incremental-scan \ + DROBOTICFLOW_OPERATOR_SLUG=bos_incremental_scan \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_source \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/bridgedata_tfds_manifest_validate/Dockerfile b/operator-images/builtin/bridgedata_tfds_manifest_validate/Dockerfile new file mode 100644 index 00000000..051f9ea8 --- /dev/null +++ b/operator-images/builtin/bridgedata_tfds_manifest_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-bridgedata-tfds-manifest-validate \ + DROBOTICFLOW_OPERATOR_SLUG=bridgedata_tfds_manifest_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_bridgedata_tfds_manifest_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/calvin_episode_validate/Dockerfile b/operator-images/builtin/calvin_episode_validate/Dockerfile new file mode 100644 index 00000000..d3e53a85 --- /dev/null +++ b/operator-images/builtin/calvin_episode_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-calvin-episode-validate \ + DROBOTICFLOW_OPERATOR_SLUG=calvin_episode_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_calvin_episode_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/camera_calibration_validate/Dockerfile b/operator-images/builtin/camera_calibration_validate/Dockerfile new file mode 100644 index 00000000..31b9e515 --- /dev/null +++ b/operator-images/builtin/camera_calibration_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-camera-calibration-validate \ + DROBOTICFLOW_OPERATOR_SLUG=camera_calibration_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_camera_calibration_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/coco_annotation_validate/Dockerfile b/operator-images/builtin/coco_annotation_validate/Dockerfile new file mode 100644 index 00000000..38d8355c --- /dev/null +++ b/operator-images/builtin/coco_annotation_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-coco-annotation-validate \ + DROBOTICFLOW_OPERATOR_SLUG=coco_annotation_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_coco_annotation_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/conditional_route/Dockerfile b/operator-images/builtin/conditional_route/Dockerfile new file mode 100644 index 00000000..079d4fe9 --- /dev/null +++ b/operator-images/builtin/conditional_route/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-conditional-route \ + DROBOTICFLOW_OPERATOR_SLUG=conditional_route \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_condition_router \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/csv_to_json/Dockerfile b/operator-images/builtin/csv_to_json/Dockerfile new file mode 100644 index 00000000..0eb95ec7 --- /dev/null +++ b/operator-images/builtin/csv_to_json/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-csv-to-json \ + DROBOTICFLOW_OPERATOR_SLUG=csv_to_json \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_csv_to_json \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/dataset_provenance_audit/Dockerfile b/operator-images/builtin/dataset_provenance_audit/Dockerfile new file mode 100644 index 00000000..ffe33a3f --- /dev/null +++ b/operator-images/builtin/dataset_provenance_audit/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-dataset-provenance-audit \ + DROBOTICFLOW_OPERATOR_SLUG=dataset_provenance_audit \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_dataset_provenance_audit \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/dataset_schema_contract_validate/Dockerfile b/operator-images/builtin/dataset_schema_contract_validate/Dockerfile new file mode 100644 index 00000000..467539bc --- /dev/null +++ b/operator-images/builtin/dataset_schema_contract_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-dataset-schema-contract-validate \ + DROBOTICFLOW_OPERATOR_SLUG=dataset_schema_contract_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_dataset_schema_contract_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/dataset_split_leakage_validate/Dockerfile b/operator-images/builtin/dataset_split_leakage_validate/Dockerfile new file mode 100644 index 00000000..62b989e3 --- /dev/null +++ b/operator-images/builtin/dataset_split_leakage_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-dataset-split-leakage-validate \ + DROBOTICFLOW_OPERATOR_SLUG=dataset_split_leakage_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_dataset_split_leakage_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/dataset_split_manifest/Dockerfile b/operator-images/builtin/dataset_split_manifest/Dockerfile new file mode 100644 index 00000000..fcd191e6 --- /dev/null +++ b/operator-images/builtin/dataset_split_manifest/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-dataset-split-manifest \ + DROBOTICFLOW_OPERATOR_SLUG=dataset_split_manifest \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_dataset_split_manifest \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/dexycb_dataset_validate/Dockerfile b/operator-images/builtin/dexycb_dataset_validate/Dockerfile new file mode 100644 index 00000000..e6effce1 --- /dev/null +++ b/operator-images/builtin/dexycb_dataset_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-dexycb-dataset-validate \ + DROBOTICFLOW_OPERATOR_SLUG=dexycb_dataset_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_dexycb_dataset_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/diffusion_policy_train/Dockerfile b/operator-images/builtin/diffusion_policy_train/Dockerfile new file mode 100644 index 00000000..e04c14de --- /dev/null +++ b/operator-images/builtin/diffusion_policy_train/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM pytorch/pytorch:2.5.1-cuda12.4-cudnn9-runtime + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-diffusion-policy-train \ + DROBOTICFLOW_OPERATOR_SLUG=diffusion_policy_train \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_diffusion_policy_train \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py torch torchvision transformers accelerate safetensors +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/directory_scan/Dockerfile b/operator-images/builtin/directory_scan/Dockerfile new file mode 100644 index 00000000..f69226c4 --- /dev/null +++ b/operator-images/builtin/directory_scan/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-directory-scan \ + DROBOTICFLOW_OPERATOR_SLUG=directory_scan \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_directory_scan \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/droid_raw_episode_validate/Dockerfile b/operator-images/builtin/droid_raw_episode_validate/Dockerfile new file mode 100644 index 00000000..41f6388f --- /dev/null +++ b/operator-images/builtin/droid_raw_episode_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-droid-raw-episode-validate \ + DROBOTICFLOW_OPERATOR_SLUG=droid_raw_episode_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_droid_raw_episode_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/embodiment_metadata_validate/Dockerfile b/operator-images/builtin/embodiment_metadata_validate/Dockerfile new file mode 100644 index 00000000..cd4598a9 --- /dev/null +++ b/operator-images/builtin/embodiment_metadata_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-embodiment-metadata-validate \ + DROBOTICFLOW_OPERATOR_SLUG=embodiment_metadata_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_embodiment_metadata_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/episode_boundary_validate/Dockerfile b/operator-images/builtin/episode_boundary_validate/Dockerfile new file mode 100644 index 00000000..40852905 --- /dev/null +++ b/operator-images/builtin/episode_boundary_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-episode-boundary-validate \ + DROBOTICFLOW_OPERATOR_SLUG=episode_boundary_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_episode_boundary_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/force_torque_sensor_validate/Dockerfile b/operator-images/builtin/force_torque_sensor_validate/Dockerfile new file mode 100644 index 00000000..09dff106 --- /dev/null +++ b/operator-images/builtin/force_torque_sensor_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-force-torque-sensor-validate \ + DROBOTICFLOW_OPERATOR_SLUG=force_torque_sensor_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_force_torque_sensor_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/furniturebench_pkl_demonstration_validate/Dockerfile b/operator-images/builtin/furniturebench_pkl_demonstration_validate/Dockerfile new file mode 100644 index 00000000..469efe51 --- /dev/null +++ b/operator-images/builtin/furniturebench_pkl_demonstration_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-furniturebench-pkl-demonstration-validate \ + DROBOTICFLOW_OPERATOR_SLUG=furniturebench_pkl_demonstration_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_furniturebench_pkl_demonstration_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/graspnet_dataset_validate/Dockerfile b/operator-images/builtin/graspnet_dataset_validate/Dockerfile new file mode 100644 index 00000000..cf440823 --- /dev/null +++ b/operator-images/builtin/graspnet_dataset_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-graspnet-dataset-validate \ + DROBOTICFLOW_OPERATOR_SLUG=graspnet_dataset_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_graspnet_dataset_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/hand_object_pose_validate/Dockerfile b/operator-images/builtin/hand_object_pose_validate/Dockerfile new file mode 100644 index 00000000..96a3966d --- /dev/null +++ b/operator-images/builtin/hand_object_pose_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-hand-object-pose-validate \ + DROBOTICFLOW_OPERATOR_SLUG=hand_object_pose_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_hand_object_pose_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/hdf5_demonstration_summary/Dockerfile b/operator-images/builtin/hdf5_demonstration_summary/Dockerfile new file mode 100644 index 00000000..7ac0c0d0 --- /dev/null +++ b/operator-images/builtin/hdf5_demonstration_summary/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-hdf5-demonstration-summary \ + DROBOTICFLOW_OPERATOR_SLUG=hdf5_demonstration_summary \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_hdf5_demonstration_summary \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/hf_dataset_probe/Dockerfile b/operator-images/builtin/hf_dataset_probe/Dockerfile new file mode 100644 index 00000000..88eeb06c --- /dev/null +++ b/operator-images/builtin/hf_dataset_probe/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-hf-dataset-probe \ + DROBOTICFLOW_OPERATOR_SLUG=hf_dataset_probe \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_hf_dataset_probe \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/ho3d_dataset_validate/Dockerfile b/operator-images/builtin/ho3d_dataset_validate/Dockerfile new file mode 100644 index 00000000..be7b0f04 --- /dev/null +++ b/operator-images/builtin/ho3d_dataset_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-ho3d-dataset-validate \ + DROBOTICFLOW_OPERATOR_SLUG=ho3d_dataset_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_ho3d_dataset_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/human_review/Dockerfile b/operator-images/builtin/human_review/Dockerfile new file mode 100644 index 00000000..3f6de8a1 --- /dev/null +++ b/operator-images/builtin/human_review/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-human-review \ + DROBOTICFLOW_OPERATOR_SLUG=human_review \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_human_review \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/joint_limit_validate/Dockerfile b/operator-images/builtin/joint_limit_validate/Dockerfile new file mode 100644 index 00000000..319d56c9 --- /dev/null +++ b/operator-images/builtin/joint_limit_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-joint-limit-validate \ + DROBOTICFLOW_OPERATOR_SLUG=joint_limit_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_joint_limit_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/json_field_extract/Dockerfile b/operator-images/builtin/json_field_extract/Dockerfile new file mode 100644 index 00000000..e9659bb6 --- /dev/null +++ b/operator-images/builtin/json_field_extract/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-json-field-extract \ + DROBOTICFLOW_OPERATOR_SLUG=json_field_extract \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_json_extract \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/json_filter/Dockerfile b/operator-images/builtin/json_filter/Dockerfile new file mode 100644 index 00000000..e9b82244 --- /dev/null +++ b/operator-images/builtin/json_filter/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-json-filter \ + DROBOTICFLOW_OPERATOR_SLUG=json_filter \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_json_filter \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/json_map_rename/Dockerfile b/operator-images/builtin/json_map_rename/Dockerfile new file mode 100644 index 00000000..619b072b --- /dev/null +++ b/operator-images/builtin/json_map_rename/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-json-map-rename \ + DROBOTICFLOW_OPERATOR_SLUG=json_map_rename \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_json_map \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/json_to_csv/Dockerfile b/operator-images/builtin/json_to_csv/Dockerfile new file mode 100644 index 00000000..70783b7a --- /dev/null +++ b/operator-images/builtin/json_to_csv/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-json-to-csv \ + DROBOTICFLOW_OPERATOR_SLUG=json_to_csv \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_json_to_csv \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/language_instruction_validate/Dockerfile b/operator-images/builtin/language_instruction_validate/Dockerfile new file mode 100644 index 00000000..32d37d52 --- /dev/null +++ b/operator-images/builtin/language_instruction_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-language-instruction-validate \ + DROBOTICFLOW_OPERATOR_SLUG=language_instruction_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_language_instruction_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/lerobot_dataset_validate/Dockerfile b/operator-images/builtin/lerobot_dataset_validate/Dockerfile new file mode 100644 index 00000000..72a9bc9f --- /dev/null +++ b/operator-images/builtin/lerobot_dataset_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-lerobot-dataset-validate \ + DROBOTICFLOW_OPERATOR_SLUG=lerobot_dataset_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_lerobot_dataset_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/lerobot_metadata_consistency_validate/Dockerfile b/operator-images/builtin/lerobot_metadata_consistency_validate/Dockerfile new file mode 100644 index 00000000..330f99e4 --- /dev/null +++ b/operator-images/builtin/lerobot_metadata_consistency_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-lerobot-metadata-consistency-validate \ + DROBOTICFLOW_OPERATOR_SLUG=lerobot_metadata_consistency_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_lerobot_metadata_consistency_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/lerobot_summary/Dockerfile b/operator-images/builtin/lerobot_summary/Dockerfile new file mode 100644 index 00000000..ab67e896 --- /dev/null +++ b/operator-images/builtin/lerobot_summary/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-lerobot-summary \ + DROBOTICFLOW_OPERATOR_SLUG=lerobot_summary \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_lerobot_summary \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/lerobot_v3_episode_index_validate/Dockerfile b/operator-images/builtin/lerobot_v3_episode_index_validate/Dockerfile new file mode 100644 index 00000000..b7867bea --- /dev/null +++ b/operator-images/builtin/lerobot_v3_episode_index_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-lerobot-v3-episode-index-validate \ + DROBOTICFLOW_OPERATOR_SLUG=lerobot_v3_episode_index_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_lerobot_v3_episode_index_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/lerobot_video_alignment_validate/Dockerfile b/operator-images/builtin/lerobot_video_alignment_validate/Dockerfile new file mode 100644 index 00000000..52a305fe --- /dev/null +++ b/operator-images/builtin/lerobot_video_alignment_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-lerobot-video-alignment-validate \ + DROBOTICFLOW_OPERATOR_SLUG=lerobot_video_alignment_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_lerobot_video_alignment_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/libero_hdf5_task_validate/Dockerfile b/operator-images/builtin/libero_hdf5_task_validate/Dockerfile new file mode 100644 index 00000000..88a38b1e --- /dev/null +++ b/operator-images/builtin/libero_hdf5_task_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-libero-hdf5-task-validate \ + DROBOTICFLOW_OPERATOR_SLUG=libero_hdf5_task_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_libero_hdf5_task_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/maniskill_trajectory_validate/Dockerfile b/operator-images/builtin/maniskill_trajectory_validate/Dockerfile new file mode 100644 index 00000000..0d9483f7 --- /dev/null +++ b/operator-images/builtin/maniskill_trajectory_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-maniskill-trajectory-validate \ + DROBOTICFLOW_OPERATOR_SLUG=maniskill_trajectory_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_maniskill_trajectory_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/mcap_topic_summary/Dockerfile b/operator-images/builtin/mcap_topic_summary/Dockerfile new file mode 100644 index 00000000..0e4f6def --- /dev/null +++ b/operator-images/builtin/mcap_topic_summary/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-mcap-topic-summary \ + DROBOTICFLOW_OPERATOR_SLUG=mcap_topic_summary \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_mcap_topic_summary \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/media_reference_validate/Dockerfile b/operator-images/builtin/media_reference_validate/Dockerfile new file mode 100644 index 00000000..70160e53 --- /dev/null +++ b/operator-images/builtin/media_reference_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-media-reference-validate \ + DROBOTICFLOW_OPERATOR_SLUG=media_reference_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_media_reference_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/merge/Dockerfile b/operator-images/builtin/merge/Dockerfile new file mode 100644 index 00000000..201d7905 --- /dev/null +++ b/operator-images/builtin/merge/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-merge \ + DROBOTICFLOW_OPERATOR_SLUG=merge \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_merge \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/mjcf_actuator_validate/Dockerfile b/operator-images/builtin/mjcf_actuator_validate/Dockerfile new file mode 100644 index 00000000..20382f98 --- /dev/null +++ b/operator-images/builtin/mjcf_actuator_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-mjcf-actuator-validate \ + DROBOTICFLOW_OPERATOR_SLUG=mjcf_actuator_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_mjcf_actuator_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/model_inference/Dockerfile b/operator-images/builtin/model_inference/Dockerfile new file mode 100644 index 00000000..1263349f --- /dev/null +++ b/operator-images/builtin/model_inference/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-model-inference \ + DROBOTICFLOW_OPERATOR_SLUG=model_inference \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_model_inference \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/mongodb_write/Dockerfile b/operator-images/builtin/mongodb_write/Dockerfile new file mode 100644 index 00000000..e1835704 --- /dev/null +++ b/operator-images/builtin/mongodb_write/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-mongodb-write \ + DROBOTICFLOW_OPERATOR_SLUG=mongodb_write \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_mongodb_write \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/openvla_finetune/Dockerfile b/operator-images/builtin/openvla_finetune/Dockerfile new file mode 100644 index 00000000..776c13b0 --- /dev/null +++ b/operator-images/builtin/openvla_finetune/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM pytorch/pytorch:2.5.1-cuda12.4-cudnn9-runtime + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-openvla-finetune \ + DROBOTICFLOW_OPERATOR_SLUG=openvla_finetune \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_openvla_finetune \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py torch torchvision transformers accelerate safetensors +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/openvla_inference/Dockerfile b/operator-images/builtin/openvla_inference/Dockerfile new file mode 100644 index 00000000..e371f767 --- /dev/null +++ b/operator-images/builtin/openvla_inference/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM pytorch/pytorch:2.5.1-cuda12.4-cudnn9-runtime + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-openvla-inference \ + DROBOTICFLOW_OPERATOR_SLUG=openvla_inference \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_openvla_inference \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py torch torchvision transformers accelerate safetensors +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/operator-images.json b/operator-images/builtin/operator-images.json new file mode 100644 index 00000000..8126eda4 --- /dev/null +++ b/operator-images/builtin/operator-images.json @@ -0,0 +1,2070 @@ +{ + "items": [ + { + "baseImage": "pytorch/pytorch:2.5.1-cuda12.4-cudnn9-runtime", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "real-act-training-command", + "trajectory-demonstration-files", + "gpu-a800" + ], + "dockerfile": "operator-images/builtin/act_policy_train/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-act-policy-train:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-act-policy-train@sha256:d24b5417b2eab1abf5bb07a1d66d38306ea5545226d3f2a3e34bb5aee184f4f2", + "operatorId": "operator-builtin-act-policy-train", + "operatorSlug": "act_policy_train", + "requiresGpu": true, + "runtimeHandler": "builtin_act_policy_train", + "runtimeSource": "backend/operators/builtin/act_policy_train/operator.py", + "sourceRoot": "backend/operators/builtin/act_policy_train" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "json-jsonl-parquet-files", + "pyarrow" + ], + "dockerfile": "operator-images/builtin/action_state_quality_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-action-state-quality-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-action-state-quality-validate@sha256:96055934c67458386848e774725ea4328cb0b1fe157e77a23e8586659e7b538d", + "operatorId": "operator-builtin-action-state-quality-validate", + "operatorSlug": "action_state_quality_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_action_state_quality_validate", + "runtimeSource": "backend/operators/builtin/action_state_quality_validate/operator.py", + "sourceRoot": "backend/operators/builtin/action_state_quality_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "agibotworld2026-meta-data-video-files", + "pyarrow", + "parquet-files" + ], + "dockerfile": "operator-images/builtin/agibot_world2026_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-agibot-world2026-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-agibot-world2026-validate@sha256:622499759d3ae4bb0bbcbd51069fb2525510f41d3a42a474e30b7d21d9df41b8", + "operatorId": "operator-builtin-agibot-world2026-validate", + "operatorSlug": "agibot_world2026_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_agibot_world2026_validate", + "runtimeSource": "backend/operators/builtin/agibot_world2026_validate/operator.py", + "sourceRoot": "backend/operators/builtin/agibot_world2026_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "aloha-act-hdf5-files", + "h5py", + "numpy" + ], + "dockerfile": "operator-images/builtin/aloha_hdf5_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-aloha-hdf5-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-aloha-hdf5-validate@sha256:e90db43b9bf900fec24f36546724121bab2a329f9ab7b48f579e84a53eb045a8", + "operatorId": "operator-builtin-aloha-hdf5-validate", + "operatorSlug": "aloha_hdf5_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_aloha_hdf5_validate", + "runtimeSource": "backend/operators/builtin/aloha_hdf5_validate/operator.py", + "sourceRoot": "backend/operators/builtin/aloha_hdf5_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "json-artifacts" + ], + "dockerfile": "operator-images/builtin/audit_sample/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-audit-sample:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-audit-sample@sha256:2a89baaae19e6c2c33980f41921020043888dc48db23cbc69c7de6f4670c9fc2", + "operatorId": "operator-builtin-audit-sample", + "operatorSlug": "audit_sample", + "requiresGpu": false, + "runtimeHandler": "builtin_audit_sampling", + "runtimeSource": "backend/operators/builtin/audit_sample/operator.py", + "sourceRoot": "backend/operators/builtin/audit_sample" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "bop-scenewise-dataset-files", + "json-files", + "rgb-depth-mask-image-files", + "ply-model-files" + ], + "dockerfile": "operator-images/builtin/bop_dataset_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-bop-dataset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-bop-dataset-validate@sha256:6b99f50e0777e02df933fb196666f6bace7c251fcdfedf1630844525905d5c1b", + "operatorId": "operator-builtin-bop-dataset-validate", + "operatorSlug": "bop_dataset_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_bop_dataset_validate", + "runtimeSource": "backend/operators/builtin/bop_dataset_validate/operator.py", + "sourceRoot": "backend/operators/builtin/bop_dataset_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "baidu-bos" + ], + "dockerfile": "operator-images/builtin/bos_incremental_scan/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-bos-incremental-scan:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-bos-incremental-scan@sha256:9077ae0f63cdf0fe76ce363ea08b84f071b2fc3c0ef516bd797a8d63eba91249", + "operatorId": "operator-builtin-bos-incremental-scan", + "operatorSlug": "bos_incremental_scan", + "requiresGpu": false, + "runtimeHandler": "builtin_source", + "runtimeSource": "backend/operators/builtin/bos_incremental_scan/operator.py", + "sourceRoot": "backend/operators/builtin/bos_incremental_scan" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "bridgedata-tfds-rlds-directory", + "tfrecord-files", + "crc32c-structural-validation" + ], + "dockerfile": "operator-images/builtin/bridgedata_tfds_manifest_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-bridgedata-tfds-manifest-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-bridgedata-tfds-manifest-validate@sha256:18c89efc04d3ca65154c07851231c16953900016178d6183041850e6e7f7e2a0", + "operatorId": "operator-builtin-bridgedata-tfds-manifest-validate", + "operatorSlug": "bridgedata_tfds_manifest_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_bridgedata_tfds_manifest_validate", + "runtimeSource": "backend/operators/builtin/bridgedata_tfds_manifest_validate/operator.py", + "sourceRoot": "backend/operators/builtin/bridgedata_tfds_manifest_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "calvin-episode-npz-files", + "numpy", + "language-annotation-files" + ], + "dockerfile": "operator-images/builtin/calvin_episode_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-calvin-episode-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-calvin-episode-validate@sha256:bc90a289bb258dc0f74fd1a33d2baa5ca53182e78eed23c42106205325ae9a48", + "operatorId": "operator-builtin-calvin-episode-validate", + "operatorSlug": "calvin_episode_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_calvin_episode_validate", + "runtimeSource": "backend/operators/builtin/calvin_episode_validate/operator.py", + "sourceRoot": "backend/operators/builtin/calvin_episode_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "camera-calibration-files", + "json-files", + "optional-yaml-files" + ], + "dockerfile": "operator-images/builtin/camera_calibration_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-camera-calibration-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-camera-calibration-validate@sha256:b43c83210cdd287d8234ff4e137586b1e5d671945ef72308246e9cceb8d7cae7", + "operatorId": "operator-builtin-camera-calibration-validate", + "operatorSlug": "camera_calibration_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_camera_calibration_validate", + "runtimeSource": "backend/operators/builtin/camera_calibration_validate/operator.py", + "sourceRoot": "backend/operators/builtin/camera_calibration_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "coco-json", + "image-files" + ], + "dockerfile": "operator-images/builtin/coco_annotation_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-coco-annotation-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-coco-annotation-validate@sha256:538e5ffdc881f6f93ae55b84c216b57954bbf662a1d1a34588ed4981eb8ab57d", + "operatorId": "operator-builtin-coco-annotation-validate", + "operatorSlug": "coco_annotation_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_coco_annotation_validate", + "runtimeSource": "backend/operators/builtin/coco_annotation_validate/operator.py", + "sourceRoot": "backend/operators/builtin/coco_annotation_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "json-artifacts" + ], + "dockerfile": "operator-images/builtin/conditional_route/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-conditional-route:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-conditional-route@sha256:0de386aacb81c0658bac316e50689d3fa56bc7403865ce3237262e1dda0ff5a4", + "operatorId": "operator-builtin-conditional-route", + "operatorSlug": "conditional_route", + "requiresGpu": false, + "runtimeHandler": "builtin_condition_router", + "runtimeSource": "backend/operators/builtin/conditional_route/operator.py", + "sourceRoot": "backend/operators/builtin/conditional_route" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "csv-files" + ], + "dockerfile": "operator-images/builtin/csv_to_json/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-csv-to-json:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-csv-to-json@sha256:9137bbc331c43d783ac20b3316c606612d96c1e725c95bfdac02f6de6d163c7f", + "operatorId": "operator-builtin-csv-to-json", + "operatorSlug": "csv_to_json", + "requiresGpu": false, + "runtimeHandler": "builtin_csv_to_json", + "runtimeSource": "backend/operators/builtin/csv_to_json/operator.py", + "sourceRoot": "backend/operators/builtin/csv_to_json" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "dataset-card-files", + "license-files", + "pyyaml" + ], + "dockerfile": "operator-images/builtin/dataset_provenance_audit/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dataset-provenance-audit:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dataset-provenance-audit@sha256:641ef1697d6e61773dc587c9142a50b920738107974f7ffbacf2512302aee98f", + "operatorId": "operator-builtin-dataset-provenance-audit", + "operatorSlug": "dataset_provenance_audit", + "requiresGpu": false, + "runtimeHandler": "builtin_dataset_provenance_audit", + "runtimeSource": "backend/operators/builtin/dataset_provenance_audit/operator.py", + "sourceRoot": "backend/operators/builtin/dataset_provenance_audit" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "csv-json-jsonl-parquet-hdf5-npz-files", + "pyarrow", + "h5py", + "numpy" + ], + "dockerfile": "operator-images/builtin/dataset_schema_contract_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dataset-schema-contract-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dataset-schema-contract-validate@sha256:9dfcb8cf18e09069e14db67c8649352071fcdf44b1d537465d38aff532f5ba48", + "operatorId": "operator-builtin-dataset-schema-contract-validate", + "operatorSlug": "dataset_schema_contract_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_dataset_schema_contract_validate", + "runtimeSource": "backend/operators/builtin/dataset_schema_contract_validate/operator.py", + "sourceRoot": "backend/operators/builtin/dataset_schema_contract_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "split-manifest-csv-json-jsonl-parquet-files", + "pyarrow" + ], + "dockerfile": "operator-images/builtin/dataset_split_leakage_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dataset-split-leakage-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dataset-split-leakage-validate@sha256:2dde8d42ac5001e4727b6061c54970c136f7588c1bde04fa7b5e3473e49aaac4", + "operatorId": "operator-builtin-dataset-split-leakage-validate", + "operatorSlug": "dataset_split_leakage_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_dataset_split_leakage_validate", + "runtimeSource": "backend/operators/builtin/dataset_split_leakage_validate/operator.py", + "sourceRoot": "backend/operators/builtin/dataset_split_leakage_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "json-artifacts", + "csv-files" + ], + "dockerfile": "operator-images/builtin/dataset_split_manifest/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dataset-split-manifest:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dataset-split-manifest@sha256:b2bcd44ccaa4b3c2b9a22d4f10a5d8101a0e19e249e693a501e1e506887fe1da", + "operatorId": "operator-builtin-dataset-split-manifest", + "operatorSlug": "dataset_split_manifest", + "requiresGpu": false, + "runtimeHandler": "builtin_dataset_split_manifest", + "runtimeSource": "backend/operators/builtin/dataset_split_manifest/operator.py", + "sourceRoot": "backend/operators/builtin/dataset_split_manifest" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "dexycb-dataset-files", + "numpy", + "npz-label-files", + "rgb-depth-image-files" + ], + "dockerfile": "operator-images/builtin/dexycb_dataset_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dexycb-dataset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-dexycb-dataset-validate@sha256:b412bc8d84fa31ff1233f13c0758156013ec03e9cd1092e9871b8bb8b4b9e6f1", + "operatorId": "operator-builtin-dexycb-dataset-validate", + "operatorSlug": "dexycb_dataset_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_dexycb_dataset_validate", + "runtimeSource": "backend/operators/builtin/dexycb_dataset_validate/operator.py", + "sourceRoot": "backend/operators/builtin/dexycb_dataset_validate" + }, + { + "baseImage": "pytorch/pytorch:2.5.1-cuda12.4-cudnn9-runtime", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "diffusion-policy-training-command", + "zarr-or-lerobot-trajectory-files", + "gpu-a800" + ], + "dockerfile": "operator-images/builtin/diffusion_policy_train/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-diffusion-policy-train:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-diffusion-policy-train@sha256:78d74d404ed7e98c19518e28315a897313c8583c6f6b0eb191e36eabcc8a2f33", + "operatorId": "operator-builtin-diffusion-policy-train", + "operatorSlug": "diffusion_policy_train", + "requiresGpu": true, + "runtimeHandler": "builtin_diffusion_policy_train", + "runtimeSource": "backend/operators/builtin/diffusion_policy_train/operator.py", + "sourceRoot": "backend/operators/builtin/diffusion_policy_train" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "filesystem" + ], + "dockerfile": "operator-images/builtin/directory_scan/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-directory-scan:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-directory-scan@sha256:f591c68e5a9f63ed1203347dedfddfe7f9f8786266d69af1fbdab74bc4eece9f", + "operatorId": "operator-builtin-directory-scan", + "operatorSlug": "directory_scan", + "requiresGpu": false, + "runtimeHandler": "builtin_directory_scan", + "runtimeSource": "backend/operators/builtin/directory_scan/operator.py", + "sourceRoot": "backend/operators/builtin/directory_scan" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "droid-raw-episode-hdf5-metadata-video-files", + "h5py", + "numpy" + ], + "dockerfile": "operator-images/builtin/droid_raw_episode_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-droid-raw-episode-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-droid-raw-episode-validate@sha256:eea450070b6b1fa4fbd80c9f1e476eacba25683791b0b40320639c65cc5fa418", + "operatorId": "operator-builtin-droid-raw-episode-validate", + "operatorSlug": "droid_raw_episode_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_droid_raw_episode_validate", + "runtimeSource": "backend/operators/builtin/droid_raw_episode_validate/operator.py", + "sourceRoot": "backend/operators/builtin/droid_raw_episode_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "csv-json-jsonl-parquet-metadata-files", + "pyarrow" + ], + "dockerfile": "operator-images/builtin/embodiment_metadata_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-embodiment-metadata-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-embodiment-metadata-validate@sha256:47df5e78b0fd62bf046f8340b7af5d4cef3bc38efaa2b10a3a7ee028adc62ebc", + "operatorId": "operator-builtin-embodiment-metadata-validate", + "operatorSlug": "embodiment_metadata_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_embodiment_metadata_validate", + "runtimeSource": "backend/operators/builtin/embodiment_metadata_validate/operator.py", + "sourceRoot": "backend/operators/builtin/embodiment_metadata_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "json-jsonl-parquet-files", + "pyarrow" + ], + "dockerfile": "operator-images/builtin/episode_boundary_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-episode-boundary-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-episode-boundary-validate@sha256:9e7782f3fe626f858138fd066ca2911d02cc4bfe6ab933065b4ebfedaff544ba", + "operatorId": "operator-builtin-episode-boundary-validate", + "operatorSlug": "episode_boundary_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_episode_boundary_validate", + "runtimeSource": "backend/operators/builtin/episode_boundary_validate/operator.py", + "sourceRoot": "backend/operators/builtin/episode_boundary_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "csv-json-jsonl-parquet-hdf5-force-torque-files", + "pyarrow", + "h5py", + "numpy" + ], + "dockerfile": "operator-images/builtin/force_torque_sensor_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-force-torque-sensor-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-force-torque-sensor-validate@sha256:d7e4558fd8fad892f328af10f21eaa8ca3f8122fd7fe209184b99c5f51a7a22a", + "operatorId": "operator-builtin-force-torque-sensor-validate", + "operatorSlug": "force_torque_sensor_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_force_torque_sensor_validate", + "runtimeSource": "backend/operators/builtin/force_torque_sensor_validate/operator.py", + "sourceRoot": "backend/operators/builtin/force_torque_sensor_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "furniturebench-pkl-files", + "pickle", + "numpy" + ], + "dockerfile": "operator-images/builtin/furniturebench_pkl_demonstration_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-furniturebench-pkl-demonstration-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-furniturebench-pkl-demonstration-validate@sha256:6633bd68e09a6a82997e46bd6f02a0a3b27f77ba40aab0094239cd11317f7c8c", + "operatorId": "operator-builtin-furniturebench-pkl-demonstration-validate", + "operatorSlug": "furniturebench_pkl_demonstration_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_furniturebench_pkl_demonstration_validate", + "runtimeSource": "backend/operators/builtin/furniturebench_pkl_demonstration_validate/operator.py", + "sourceRoot": "backend/operators/builtin/furniturebench_pkl_demonstration_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "graspnet-1billion-dataset-files", + "numpy", + "npz-npy-files", + "rgb-depth-mask-png-files", + "xml-annotations" + ], + "dockerfile": "operator-images/builtin/graspnet_dataset_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-graspnet-dataset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-graspnet-dataset-validate@sha256:c4ec92cc4e87e5df975377608f02ba54de0570d0235dd36840d4a311875c060a", + "operatorId": "operator-builtin-graspnet-dataset-validate", + "operatorSlug": "graspnet_dataset_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_graspnet_dataset_validate", + "runtimeSource": "backend/operators/builtin/graspnet_dataset_validate/operator.py", + "sourceRoot": "backend/operators/builtin/graspnet_dataset_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "json-jsonl-csv-parquet-files", + "image-files", + "mask-files", + "pyarrow" + ], + "dockerfile": "operator-images/builtin/hand_object_pose_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-hand-object-pose-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-hand-object-pose-validate@sha256:f91c6c745f098d7a592c3738b4e9bbb4a2be1604605da5ccd539ef90b010deed", + "operatorId": "operator-builtin-hand-object-pose-validate", + "operatorSlug": "hand_object_pose_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_hand_object_pose_validate", + "runtimeSource": "backend/operators/builtin/hand_object_pose_validate/operator.py", + "sourceRoot": "backend/operators/builtin/hand_object_pose_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "hdf5-files", + "h5py" + ], + "dockerfile": "operator-images/builtin/hdf5_demonstration_summary/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-hdf5-demonstration-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-hdf5-demonstration-summary@sha256:3d82e565a006c7c300d45ebfffaa1098d1d6b3e8bcdffd1b8a361d0768b7e08f", + "operatorId": "operator-builtin-hdf5-demonstration-summary", + "operatorSlug": "hdf5_demonstration_summary", + "requiresGpu": false, + "runtimeHandler": "builtin_hdf5_demonstration_summary", + "runtimeSource": "backend/operators/builtin/hdf5_demonstration_summary/operator.py", + "sourceRoot": "backend/operators/builtin/hdf5_demonstration_summary" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "huggingface-dataset-viewer-api" + ], + "dockerfile": "operator-images/builtin/hf_dataset_probe/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-hf-dataset-probe:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-hf-dataset-probe@sha256:8ec0202d73059352be727b28c12daf9a6bdfd77e7e9ea3332d658efde66654c4", + "operatorId": "operator-builtin-hf-dataset-probe", + "operatorSlug": "hf_dataset_probe", + "requiresGpu": false, + "runtimeHandler": "builtin_hf_dataset_probe", + "runtimeSource": "backend/operators/builtin/hf_dataset_probe/operator.py", + "sourceRoot": "backend/operators/builtin/hf_dataset_probe" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "ho3d-h2o3d-dataset-files", + "pickle", + "numpy", + "rgb-depth-image-files" + ], + "dockerfile": "operator-images/builtin/ho3d_dataset_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-ho3d-dataset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-ho3d-dataset-validate@sha256:9edf87bceb97b38907f6270835c2b9fcd6dd347d800efaf41a6d747eddd8af8c", + "operatorId": "operator-builtin-ho3d-dataset-validate", + "operatorSlug": "ho3d_dataset_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_ho3d_dataset_validate", + "runtimeSource": "backend/operators/builtin/ho3d_dataset_validate/operator.py", + "sourceRoot": "backend/operators/builtin/ho3d_dataset_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "json-artifacts" + ], + "dockerfile": "operator-images/builtin/human_review/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-human-review:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-human-review@sha256:8e789d73f7933102ce12364c2a382affaa16639491393abfdd3092bc3090215d", + "operatorId": "operator-builtin-human-review", + "operatorSlug": "human_review", + "requiresGpu": false, + "runtimeHandler": "builtin_human_review", + "runtimeSource": "backend/operators/builtin/human_review/operator.py", + "sourceRoot": "backend/operators/builtin/human_review" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "urdf-files", + "json-jsonl-csv-parquet-files", + "pyarrow" + ], + "dockerfile": "operator-images/builtin/joint_limit_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-joint-limit-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-joint-limit-validate@sha256:b4edf3d011f7234d3479635a6aecf33c695674827487d4c8578cb02446785b82", + "operatorId": "operator-builtin-joint-limit-validate", + "operatorSlug": "joint_limit_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_joint_limit_validate", + "runtimeSource": "backend/operators/builtin/joint_limit_validate/operator.py", + "sourceRoot": "backend/operators/builtin/joint_limit_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "json-artifacts" + ], + "dockerfile": "operator-images/builtin/json_field_extract/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-json-field-extract:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-json-field-extract@sha256:f7b6df195863eef04846a3372f02cb5cac149052496976c65891d7cb3f844945", + "operatorId": "operator-builtin-json-field-extract", + "operatorSlug": "json_field_extract", + "requiresGpu": false, + "runtimeHandler": "builtin_json_extract", + "runtimeSource": "backend/operators/builtin/json_field_extract/operator.py", + "sourceRoot": "backend/operators/builtin/json_field_extract" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "json-artifacts" + ], + "dockerfile": "operator-images/builtin/json_filter/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-json-filter:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-json-filter@sha256:824f00ea594090923c52d78d0b649abef1b47d37e0d8f1211734c39c66943bbb", + "operatorId": "operator-builtin-json-filter", + "operatorSlug": "json_filter", + "requiresGpu": false, + "runtimeHandler": "builtin_json_filter", + "runtimeSource": "backend/operators/builtin/json_filter/operator.py", + "sourceRoot": "backend/operators/builtin/json_filter" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "json-artifacts" + ], + "dockerfile": "operator-images/builtin/json_map_rename/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-json-map-rename:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-json-map-rename@sha256:53a425b6fabacbf148a17aaf411ce6762c9c5dcd9f4bdf2802edd9f12fc18c72", + "operatorId": "operator-builtin-json-map-rename", + "operatorSlug": "json_map_rename", + "requiresGpu": false, + "runtimeHandler": "builtin_json_map", + "runtimeSource": "backend/operators/builtin/json_map_rename/operator.py", + "sourceRoot": "backend/operators/builtin/json_map_rename" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "json-artifacts" + ], + "dockerfile": "operator-images/builtin/json_to_csv/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-json-to-csv:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-json-to-csv@sha256:17d52fd86f00c899e777859a4eef6fb4e16496cbdadbf085ec51c5e79b8bb724", + "operatorId": "operator-builtin-json-to-csv", + "operatorSlug": "json_to_csv", + "requiresGpu": false, + "runtimeHandler": "builtin_json_to_csv", + "runtimeSource": "backend/operators/builtin/json_to_csv/operator.py", + "sourceRoot": "backend/operators/builtin/json_to_csv" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "json-jsonl-csv-parquet-files", + "pyarrow" + ], + "dockerfile": "operator-images/builtin/language_instruction_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-language-instruction-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-language-instruction-validate@sha256:302e6fee95108b952abdbd080e18641c08717983a0fa6b2efdcf387ba25e2249", + "operatorId": "operator-builtin-language-instruction-validate", + "operatorSlug": "language_instruction_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_language_instruction_validate", + "runtimeSource": "backend/operators/builtin/language_instruction_validate/operator.py", + "sourceRoot": "backend/operators/builtin/language_instruction_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "lerobot-dataset-files", + "pyarrow", + "parquet-files" + ], + "dockerfile": "operator-images/builtin/lerobot_dataset_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-dataset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-dataset-validate@sha256:f2471c2ea75c05059fd995667b337681f8f6ff3a8947c5b960f03fae93a2d087", + "operatorId": "operator-builtin-lerobot-dataset-validate", + "operatorSlug": "lerobot_dataset_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_lerobot_dataset_validate", + "runtimeSource": "backend/operators/builtin/lerobot_dataset_validate/operator.py", + "sourceRoot": "backend/operators/builtin/lerobot_dataset_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "lerobot-v3-meta-files", + "pyarrow", + "parquet-files" + ], + "dockerfile": "operator-images/builtin/lerobot_metadata_consistency_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-metadata-consistency-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-metadata-consistency-validate@sha256:a58daedea8b3996999c73020a68a07f75c1bdbbcdb6189f5d8183c322a0d42c6", + "operatorId": "operator-builtin-lerobot-metadata-consistency-validate", + "operatorSlug": "lerobot_metadata_consistency_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_lerobot_metadata_consistency_validate", + "runtimeSource": "backend/operators/builtin/lerobot_metadata_consistency_validate/operator.py", + "sourceRoot": "backend/operators/builtin/lerobot_metadata_consistency_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "lerobot-dataset-files" + ], + "dockerfile": "operator-images/builtin/lerobot_summary/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-summary@sha256:c8e1878dced01da3da3a068cd53a395b03061b27f3e52e20cef2faedf4081dbf", + "operatorId": "operator-builtin-lerobot-summary", + "operatorSlug": "lerobot_summary", + "requiresGpu": false, + "runtimeHandler": "builtin_lerobot_summary", + "runtimeSource": "backend/operators/builtin/lerobot_summary/operator.py", + "sourceRoot": "backend/operators/builtin/lerobot_summary" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "lerobot-v3-meta-data-files", + "pyarrow", + "parquet-files" + ], + "dockerfile": "operator-images/builtin/lerobot_v3_episode_index_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-v3-episode-index-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-v3-episode-index-validate@sha256:204d8dc1e1530c8096c7aca6817ce41b12ed2c70fcd0d31fe2346427b0833456", + "operatorId": "operator-builtin-lerobot-v3-episode-index-validate", + "operatorSlug": "lerobot_v3_episode_index_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_lerobot_v3_episode_index_validate", + "runtimeSource": "backend/operators/builtin/lerobot_v3_episode_index_validate/operator.py", + "sourceRoot": "backend/operators/builtin/lerobot_v3_episode_index_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "lerobot-v3-data-files", + "pyarrow", + "parquet-files", + "video-files", + "opencv-python-headless" + ], + "dockerfile": "operator-images/builtin/lerobot_video_alignment_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-video-alignment-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-lerobot-video-alignment-validate@sha256:51022c000aed86bda84f3f33453984ed85620b87ea2487ed451864abcef7282a", + "operatorId": "operator-builtin-lerobot-video-alignment-validate", + "operatorSlug": "lerobot_video_alignment_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_lerobot_video_alignment_validate", + "runtimeSource": "backend/operators/builtin/lerobot_video_alignment_validate/operator.py", + "sourceRoot": "backend/operators/builtin/lerobot_video_alignment_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "libero-hdf5-demo-files", + "libero-bddl-files", + "h5py", + "numpy" + ], + "dockerfile": "operator-images/builtin/libero_hdf5_task_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-libero-hdf5-task-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-libero-hdf5-task-validate@sha256:68b8f8d5b203a0c644d8cec121c6d802ec55cf89ac0fa4aa87eb471b7076e18d", + "operatorId": "operator-builtin-libero-hdf5-task-validate", + "operatorSlug": "libero_hdf5_task_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_libero_hdf5_task_validate", + "runtimeSource": "backend/operators/builtin/libero_hdf5_task_validate/operator.py", + "sourceRoot": "backend/operators/builtin/libero_hdf5_task_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "maniskill-hdf5-json-files", + "h5py", + "numpy" + ], + "dockerfile": "operator-images/builtin/maniskill_trajectory_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-maniskill-trajectory-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-maniskill-trajectory-validate@sha256:d18459e85b43df78ed7df591c2ecfe86122ecb927f429cf65b465c5096edf453", + "operatorId": "operator-builtin-maniskill-trajectory-validate", + "operatorSlug": "maniskill_trajectory_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_maniskill_trajectory_validate", + "runtimeSource": "backend/operators/builtin/maniskill_trajectory_validate/operator.py", + "sourceRoot": "backend/operators/builtin/maniskill_trajectory_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "mcap-files", + "mcap-python" + ], + "dockerfile": "operator-images/builtin/mcap_topic_summary/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-mcap-topic-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-mcap-topic-summary@sha256:421e060f3d74ac067b895d83a4c7478c43feca5a75d91623859b5a1d867a5ecc", + "operatorId": "operator-builtin-mcap-topic-summary", + "operatorSlug": "mcap_topic_summary", + "requiresGpu": false, + "runtimeHandler": "builtin_mcap_topic_summary", + "runtimeSource": "backend/operators/builtin/mcap_topic_summary/operator.py", + "sourceRoot": "backend/operators/builtin/mcap_topic_summary" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "csv-json-jsonl-parquet-manifests", + "image-video-depth-pointcloud-files", + "opencv-python-headless", + "numpy", + "pyarrow", + "ffprobe" + ], + "dockerfile": "operator-images/builtin/media_reference_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-media-reference-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-media-reference-validate@sha256:60bdd6320deaf66576927bf26b25809911ea98210e9c2daa0665fe9846157104", + "operatorId": "operator-builtin-media-reference-validate", + "operatorSlug": "media_reference_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_media_reference_validate", + "runtimeSource": "backend/operators/builtin/media_reference_validate/operator.py", + "sourceRoot": "backend/operators/builtin/media_reference_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "json-artifacts" + ], + "dockerfile": "operator-images/builtin/merge/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-merge:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-merge@sha256:43dd9f250d94e9dcc2af193aa03a602a3f27ce4d3b119f5063760638cd1db0d9", + "operatorId": "operator-builtin-merge", + "operatorSlug": "merge", + "requiresGpu": false, + "runtimeHandler": "builtin_merge", + "runtimeSource": "backend/operators/builtin/merge/operator.py", + "sourceRoot": "backend/operators/builtin/merge" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "mjcf-files", + "xml-parser" + ], + "dockerfile": "operator-images/builtin/mjcf_actuator_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-mjcf-actuator-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-mjcf-actuator-validate@sha256:adc49f2a4cd9a27f2e74c74852c8a74eb33233e98973e80507346260e2abed8c", + "operatorId": "operator-builtin-mjcf-actuator-validate", + "operatorSlug": "mjcf_actuator_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_mjcf_actuator_validate", + "runtimeSource": "backend/operators/builtin/mjcf_actuator_validate/operator.py", + "sourceRoot": "backend/operators/builtin/mjcf_actuator_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "json-model-or-command" + ], + "dockerfile": "operator-images/builtin/model_inference/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-model-inference:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-model-inference@sha256:a31cf2777cda7b213f4b3fa7e004ed5ddcd31fe5f3503c5cbb039be1f6bfaaa4", + "operatorId": "operator-builtin-model-inference", + "operatorSlug": "model_inference", + "requiresGpu": false, + "runtimeHandler": "builtin_model_inference", + "runtimeSource": "backend/operators/builtin/model_inference/operator.py", + "sourceRoot": "backend/operators/builtin/model_inference" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "mongodb" + ], + "dockerfile": "operator-images/builtin/mongodb_write/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-mongodb-write:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-mongodb-write@sha256:36677374caf4ee70072c0c6ec565498565d2b667c8fb238c9ebb81bf6ecee876", + "operatorId": "operator-builtin-mongodb-write", + "operatorSlug": "mongodb_write", + "requiresGpu": false, + "runtimeHandler": "builtin_mongodb_write", + "runtimeSource": "backend/operators/builtin/mongodb_write/operator.py", + "sourceRoot": "backend/operators/builtin/mongodb_write" + }, + { + "baseImage": "pytorch/pytorch:2.5.1-cuda12.4-cudnn9-runtime", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "openvla-training-command", + "vla-sample-action-manifest", + "gpu-a800" + ], + "dockerfile": "operator-images/builtin/openvla_finetune/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-openvla-finetune:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-openvla-finetune@sha256:36184d088fb2ca79a69c6c11adfba4099a1afa32eac7ec3da6408946044837d2", + "operatorId": "operator-builtin-openvla-finetune", + "operatorSlug": "openvla_finetune", + "requiresGpu": true, + "runtimeHandler": "builtin_openvla_finetune", + "runtimeSource": "backend/operators/builtin/openvla_finetune/operator.py", + "sourceRoot": "backend/operators/builtin/openvla_finetune" + }, + { + "baseImage": "pytorch/pytorch:2.5.1-cuda12.4-cudnn9-runtime", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "openvla-inference-command", + "openvla-model-weights", + "gpu-a800" + ], + "dockerfile": "operator-images/builtin/openvla_inference/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-openvla-inference:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-openvla-inference@sha256:3336a89556d87b18a268be0234ff95acd6b23405e90c0da6fe9cdcbf7d3c524e", + "operatorId": "operator-builtin-openvla-inference", + "operatorSlug": "openvla_inference", + "requiresGpu": true, + "runtimeHandler": "builtin_openvla_inference", + "runtimeSource": "backend/operators/builtin/openvla_inference/operator.py", + "sourceRoot": "backend/operators/builtin/openvla_inference" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "filesystem" + ], + "dockerfile": "operator-images/builtin/package/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-package:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-package@sha256:e700d9d957364a1446c34f797fbdddd1c9e488c9e866913aeb740cb9b0a03741", + "operatorId": "operator-builtin-package", + "operatorSlug": "package", + "requiresGpu": false, + "runtimeHandler": "builtin_result_package", + "runtimeSource": "backend/operators/builtin/package/operator.py", + "sourceRoot": "backend/operators/builtin/package" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "parquet-files", + "pyarrow" + ], + "dockerfile": "operator-images/builtin/parquet_table_summary/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-parquet-table-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-parquet-table-summary@sha256:03e9c35aa6e3658bddd4821d6125b5bff10b4d412d33d655dfd869cdae3d03fb", + "operatorId": "operator-builtin-parquet-table-summary", + "operatorSlug": "parquet_table_summary", + "requiresGpu": false, + "runtimeHandler": "builtin_parquet_table_summary", + "runtimeSource": "backend/operators/builtin/parquet_table_summary/operator.py", + "sourceRoot": "backend/operators/builtin/parquet_table_summary" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "partnet-mobility-sapien-assets", + "urdf-files", + "json-files", + "semantics-txt", + "mesh-obj-files" + ], + "dockerfile": "operator-images/builtin/partnet_mobility_asset_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-partnet-mobility-asset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-partnet-mobility-asset-validate@sha256:52bb4ada32074e453f43aaf5fb9f7856a71eeb4e8ae6f2f0751943da55af93a0", + "operatorId": "operator-builtin-partnet-mobility-asset-validate", + "operatorSlug": "partnet_mobility_asset_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_partnet_mobility_asset_validate", + "runtimeSource": "backend/operators/builtin/partnet_mobility_asset_validate/operator.py", + "sourceRoot": "backend/operators/builtin/partnet_mobility_asset_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "pcd-files", + "ply-files" + ], + "dockerfile": "operator-images/builtin/pointcloud_summary/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-pointcloud-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-pointcloud-summary@sha256:fa846da15ab5d4816d78375fc700873996fa0ca3a0aaeb9711f106e348146235", + "operatorId": "operator-builtin-pointcloud-summary", + "operatorSlug": "pointcloud_summary", + "requiresGpu": false, + "runtimeHandler": "builtin_pointcloud_summary", + "runtimeSource": "backend/operators/builtin/pointcloud_summary/operator.py", + "sourceRoot": "backend/operators/builtin/pointcloud_summary" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "filesystem" + ], + "dockerfile": "operator-images/builtin/real_source/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-real-source:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-real-source@sha256:833d6030b8c10cabfddb3cad54137bb08f232f2bf5071863021276e6b6311d66", + "operatorId": "operator-builtin-real-source", + "operatorSlug": "real_source", + "requiresGpu": false, + "runtimeHandler": "builtin_source", + "runtimeSource": "backend/operators/builtin/real_source/operator.py", + "sourceRoot": "backend/operators/builtin/real_source" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "rgb-images", + "depth-images", + "opencv-python-headless", + "numpy", + "pyyaml" + ], + "dockerfile": "operator-images/builtin/rgbd_frame_pair_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rgbd-frame-pair-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rgbd-frame-pair-validate@sha256:4c43bb4d7b81eea7604f3dce8d074201de8187aa2b12cce1e2f7fa1f049838b8", + "operatorId": "operator-builtin-rgbd-frame-pair-validate", + "operatorSlug": "rgbd_frame_pair_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_rgbd_frame_pair_validate", + "runtimeSource": "backend/operators/builtin/rgbd_frame_pair_validate/operator.py", + "sourceRoot": "backend/operators/builtin/rgbd_frame_pair_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "rh20t-scene-directory", + "numpy", + "npy-files", + "image-video-files" + ], + "dockerfile": "operator-images/builtin/rh20t_scene_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rh20t-scene-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rh20t-scene-validate@sha256:ba1e0190c4d8283480bbed515f6a3ba26132bfc4fb74ec701b137b043f194c94", + "operatorId": "operator-builtin-rh20t-scene-validate", + "operatorSlug": "rh20t_scene_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_rh20t_scene_validate", + "runtimeSource": "backend/operators/builtin/rh20t_scene_validate/operator.py", + "sourceRoot": "backend/operators/builtin/rh20t_scene_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "rlbench-episode-files", + "pickle-files", + "image-files" + ], + "dockerfile": "operator-images/builtin/rlbench_dataset_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rlbench-dataset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rlbench-dataset-validate@sha256:f33e00d9a03bad99b9abeac51c69bf561b766b64188c744a154086364bc75764", + "operatorId": "operator-builtin-rlbench-dataset-validate", + "operatorSlug": "rlbench_dataset_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_rlbench_dataset_validate", + "runtimeSource": "backend/operators/builtin/rlbench_dataset_validate/operator.py", + "sourceRoot": "backend/operators/builtin/rlbench_dataset_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "rlds-json-or-tfds-files" + ], + "dockerfile": "operator-images/builtin/rlds_episode_summary/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rlds-episode-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rlds-episode-summary@sha256:49aae340f8d0d11ebdf89f9a01946477472c9b09b34e7f51fdef51c2feffabcb", + "operatorId": "operator-builtin-rlds-episode-summary", + "operatorSlug": "rlds_episode_summary", + "requiresGpu": false, + "runtimeHandler": "builtin_rlds_episode_summary", + "runtimeSource": "backend/operators/builtin/rlds_episode_summary/operator.py", + "sourceRoot": "backend/operators/builtin/rlds_episode_summary" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "robocasa-hdf5-demo-files", + "h5py", + "numpy" + ], + "dockerfile": "operator-images/builtin/robocasa_hdf5_demonstration_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-robocasa-hdf5-demonstration-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-robocasa-hdf5-demonstration-validate@sha256:d6010ee3a8123e3b67d406742ac822911ea556f69320ef5603edba24a3b691d5", + "operatorId": "operator-builtin-robocasa-hdf5-demonstration-validate", + "operatorSlug": "robocasa_hdf5_demonstration_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_robocasa_hdf5_demonstration_validate", + "runtimeSource": "backend/operators/builtin/robocasa_hdf5_demonstration_validate/operator.py", + "sourceRoot": "backend/operators/builtin/robocasa_hdf5_demonstration_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "robomimic-hdf5-files", + "h5py", + "numpy" + ], + "dockerfile": "operator-images/builtin/robomimic_hdf5_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-robomimic-hdf5-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-robomimic-hdf5-validate@sha256:0d6c7b88a5c3139f3087deb754dbc7642dcbcd2899c1c708a21b1f7c21a04e66", + "operatorId": "operator-builtin-robomimic-hdf5-validate", + "operatorSlug": "robomimic_hdf5_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_robomimic_hdf5_validate", + "runtimeSource": "backend/operators/builtin/robomimic_hdf5_validate/operator.py", + "sourceRoot": "backend/operators/builtin/robomimic_hdf5_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "robomind-trajectory-hdf5-files", + "h5py", + "numpy" + ], + "dockerfile": "operator-images/builtin/robomind_hdf5_quality_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-robomind-hdf5-quality-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-robomind-hdf5-quality-validate@sha256:e0973f8631bf7234ab5a0073f50f3fb79fc1cf333d0e8060596c218168f2913b", + "operatorId": "operator-builtin-robomind-hdf5-quality-validate", + "operatorSlug": "robomind_hdf5_quality_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_robomind_hdf5_quality_validate", + "runtimeSource": "backend/operators/builtin/robomind_hdf5_quality_validate/operator.py", + "sourceRoot": "backend/operators/builtin/robomind_hdf5_quality_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "roboset-robohive-hdf5-trajectory-files", + "h5py", + "numpy" + ], + "dockerfile": "operator-images/builtin/roboset_hdf5_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-roboset-hdf5-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-roboset-hdf5-validate@sha256:db36835317e9ce8c8fa5d550ddf8327460f2a891bf57a2219bd874e6eb154ec1", + "operatorId": "operator-builtin-roboset-hdf5-validate", + "operatorSlug": "roboset_hdf5_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_roboset_hdf5_validate", + "runtimeSource": "backend/operators/builtin/roboset_hdf5_validate/operator.py", + "sourceRoot": "backend/operators/builtin/roboset_hdf5_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "urdf-files", + "mjcf-files", + "mesh-assets" + ], + "dockerfile": "operator-images/builtin/robot_model_asset_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-robot-model-asset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-robot-model-asset-validate@sha256:9e85004992d213c08d88dba1c4caf0221e56c6feee9e6e92ab48ce4b4c79fe93", + "operatorId": "operator-builtin-robot-model-asset-validate", + "operatorSlug": "robot_model_asset_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_robot_model_asset_validate", + "runtimeSource": "backend/operators/builtin/robot_model_asset_validate/operator.py", + "sourceRoot": "backend/operators/builtin/robot_model_asset_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "roboturk-real-aligned-hdf5-files", + "h5py", + "numpy" + ], + "dockerfile": "operator-images/builtin/roboturk_real_hdf5_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-roboturk-real-hdf5-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-roboturk-real-hdf5-validate@sha256:db43d615bf52f8ea734d8a45e001602e91b32c7b2046e51cade16489faf4c838", + "operatorId": "operator-builtin-roboturk-real-hdf5-validate", + "operatorSlug": "roboturk_real_hdf5_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_roboturk_real_hdf5_validate", + "runtimeSource": "backend/operators/builtin/roboturk_real_hdf5_validate/operator.py", + "sourceRoot": "backend/operators/builtin/roboturk_real_hdf5_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "ros-joint-state-files", + "json-jsonl-csv-parquet-files", + "pyarrow" + ], + "dockerfile": "operator-images/builtin/ros_joint_state_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-ros-joint-state-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-ros-joint-state-validate@sha256:55199b510c1c2329facf0a12214c17fb3c4d7c256e89b6772c355a072bd878f7", + "operatorId": "operator-builtin-ros-joint-state-validate", + "operatorSlug": "ros_joint_state_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_ros_joint_state_validate", + "runtimeSource": "backend/operators/builtin/ros_joint_state_validate/operator.py", + "sourceRoot": "backend/operators/builtin/ros_joint_state_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "rosbag2-files", + "sqlite" + ], + "dockerfile": "operator-images/builtin/rosbag_summary/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rosbag-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rosbag-summary@sha256:9d41ed0073c84068869b7c844cf2a5362601dab551835a0ce0a3b1a9d00de9c5", + "operatorId": "operator-builtin-rosbag-summary", + "operatorSlug": "rosbag_summary", + "requiresGpu": false, + "runtimeHandler": "builtin_rosbag_summary", + "runtimeSource": "backend/operators/builtin/rosbag_summary/operator.py", + "sourceRoot": "backend/operators/builtin/rosbag_summary" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "csv-json-jsonl-parquet-hdf5-action-files", + "pyarrow", + "h5py", + "numpy" + ], + "dockerfile": "operator-images/builtin/rtx_action_schema_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rtx-action-schema-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-rtx-action-schema-validate@sha256:65dc02ffb329e3df1a0cadf66dc92379c92b3e98b6d35bfe460f6ea893aea031", + "operatorId": "operator-builtin-rtx-action-schema-validate", + "operatorSlug": "rtx_action_schema_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_rtx_action_schema_validate", + "runtimeSource": "backend/operators/builtin/rtx_action_schema_validate/operator.py", + "sourceRoot": "backend/operators/builtin/rtx_action_schema_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "segmentation-mask-images", + "opencv-python-headless", + "numpy" + ], + "dockerfile": "operator-images/builtin/segmentation_mask_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-segmentation-mask-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-segmentation-mask-validate@sha256:c96e3c47fe7bab282925c75d473fd1e8c400605c5d33b9dac9ab68ca98e4fbd3", + "operatorId": "operator-builtin-segmentation-mask-validate", + "operatorSlug": "segmentation_mask_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_segmentation_mask_validate", + "runtimeSource": "backend/operators/builtin/segmentation_mask_validate/operator.py", + "sourceRoot": "backend/operators/builtin/segmentation_mask_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "timestamp-csv-jsonl" + ], + "dockerfile": "operator-images/builtin/sensor_timestamp_sync_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-sensor-timestamp-sync-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-sensor-timestamp-sync-validate@sha256:4afd1efaf170d73917f215fbfdef55f36869726a32a07e79b9037204febe9a8b", + "operatorId": "operator-builtin-sensor-timestamp-sync-validate", + "operatorSlug": "sensor_timestamp_sync_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_sensor_timestamp_sync_validate", + "runtimeSource": "backend/operators/builtin/sensor_timestamp_sync_validate/operator.py", + "sourceRoot": "backend/operators/builtin/sensor_timestamp_sync_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "container-shell" + ], + "dockerfile": "operator-images/builtin/shell_command/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-shell-command:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-shell-command@sha256:47f7039f4ddcf58fb51cf38f1e4fe8fd7c96e8bcad2ea745c88ee0d9e1423f51", + "operatorId": "operator-builtin-shell-command", + "operatorSlug": "shell_command", + "requiresGpu": false, + "runtimeHandler": "builtin_shell_command", + "runtimeSource": "backend/operators/builtin/shell_command/operator.py", + "sourceRoot": "backend/operators/builtin/shell_command" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "json-artifacts", + "obj-mtl-gltf-files" + ], + "dockerfile": "operator-images/builtin/sim3d_asset_generate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-sim3d-asset-generate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-sim3d-asset-generate@sha256:c56c68886041a19492d9427211dc4741574a95eacc3242d4f34c1adb6d2bcc36", + "operatorId": "operator-builtin-sim3d-asset-generate", + "operatorSlug": "sim3d_asset_generate", + "requiresGpu": false, + "runtimeHandler": "builtin_sim3d_asset_generate", + "runtimeSource": "backend/operators/builtin/sim3d_asset_generate/operator.py", + "sourceRoot": "backend/operators/builtin/sim3d_asset_generate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "json-artifacts" + ], + "dockerfile": "operator-images/builtin/template_render/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-template-render:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-template-render@sha256:b96fcf95d3500eb6849cdb61df7a5ae6f4d3266fc5e69d6b9cce6ce2a05c8fc4", + "operatorId": "operator-builtin-template-render", + "operatorSlug": "template_render", + "requiresGpu": false, + "runtimeHandler": "builtin_text_template", + "runtimeSource": "backend/operators/builtin/template_render/operator.py", + "sourceRoot": "backend/operators/builtin/template_render" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "json-jsonl-csv-yaml-files", + "optional-pyyaml" + ], + "dockerfile": "operator-images/builtin/tf_transform_tree_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-tf-transform-tree-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-tf-transform-tree-validate@sha256:4777877db8114d07f770211c4d9b3375300bcbf52823d0abd5696fd2747bb4bc", + "operatorId": "operator-builtin-tf-transform-tree-validate", + "operatorSlug": "tf_transform_tree_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_tf_transform_tree_validate", + "runtimeSource": "backend/operators/builtin/tf_transform_tree_validate/operator.py", + "sourceRoot": "backend/operators/builtin/tf_transform_tree_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "tfrecord-files", + "crc32c-structural-validation" + ], + "dockerfile": "operator-images/builtin/tfrecord_structural_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-tfrecord-structural-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-tfrecord-structural-validate@sha256:60fb1eb05a35a6b7712c5e07ce067962ed125ef3f3e8b73853c56c5d250e30ad", + "operatorId": "operator-builtin-tfrecord-structural-validate", + "operatorSlug": "tfrecord_structural_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_tfrecord_structural_validate", + "runtimeSource": "backend/operators/builtin/tfrecord_structural_validate/operator.py", + "sourceRoot": "backend/operators/builtin/tfrecord_structural_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "trajectory-json", + "local-media-files" + ], + "dockerfile": "operator-images/builtin/trajectory_integrity_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-trajectory-integrity-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-trajectory-integrity-validate@sha256:1ec455b21a6a4cf3c35953d50d8965c77deeffd71412ffa1ebb3bc66f42fbe9e", + "operatorId": "operator-builtin-trajectory-integrity-validate", + "operatorSlug": "trajectory_integrity_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_trajectory_integrity_validate", + "runtimeSource": "backend/operators/builtin/trajectory_integrity_validate/operator.py", + "sourceRoot": "backend/operators/builtin/trajectory_integrity_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "umi-zarr-dataset", + "zarr", + "numpy" + ], + "dockerfile": "operator-images/builtin/umi_zarr_dataset_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-umi-zarr-dataset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-umi-zarr-dataset-validate@sha256:a6f4a48ad0f019a3720905ccbc5b8ab733baf6cf03cb0163695205d792831ff1", + "operatorId": "operator-builtin-umi-zarr-dataset-validate", + "operatorSlug": "umi_zarr_dataset_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_umi_zarr_dataset_validate", + "runtimeSource": "backend/operators/builtin/umi_zarr_dataset_validate/operator.py", + "sourceRoot": "backend/operators/builtin/umi_zarr_dataset_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "filesystem", + "optional-baidu-bos" + ], + "dockerfile": "operator-images/builtin/upload/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-upload:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-upload@sha256:bc56ba702686c5ed78cc2ab77ede4fa2476acba81df1aa1bd3ea5d6c7ef88474", + "operatorId": "operator-builtin-upload", + "operatorSlug": "upload", + "requiresGpu": false, + "runtimeHandler": "builtin_upload", + "runtimeSource": "backend/operators/builtin/upload/operator.py", + "sourceRoot": "backend/operators/builtin/upload" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "video-files", + "json-artifacts" + ], + "dockerfile": "operator-images/builtin/video_annotate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-annotate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-annotate@sha256:b2bb1476b0994f852a057c1b81bc412ed25c257f3581b720752c19435045f06d", + "operatorId": "operator-builtin-video-annotate", + "operatorSlug": "video_annotate", + "requiresGpu": false, + "runtimeHandler": "builtin_video_annotation", + "runtimeSource": "backend/operators/builtin/video_annotate/operator.py", + "sourceRoot": "backend/operators/builtin/video_annotate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "json-artifacts" + ], + "dockerfile": "operator-images/builtin/video_filter/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-filter:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-filter@sha256:421e6b00389e9761b1bec6b316dec38151200f055ff9bd6deb6659ce170f97ae", + "operatorId": "operator-builtin-video-filter", + "operatorSlug": "video_filter", + "requiresGpu": false, + "runtimeHandler": "builtin_dataset_filter", + "runtimeSource": "backend/operators/builtin/video_filter/operator.py", + "sourceRoot": "backend/operators/builtin/video_filter" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "ffprobe", + "video-files" + ], + "dockerfile": "operator-images/builtin/video_quality/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-quality:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-quality@sha256:755209a0ec2c869a9784e70fa4cad0803dda72f031791f1e43363bc38b3c4ece", + "operatorId": "operator-builtin-video-quality", + "operatorSlug": "video_quality", + "requiresGpu": false, + "runtimeHandler": "builtin_video_quality", + "runtimeSource": "backend/operators/builtin/video_quality/operator.py", + "sourceRoot": "backend/operators/builtin/video_quality" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "ffmpeg", + "video-files" + ], + "dockerfile": "operator-images/builtin/video_slice/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-slice:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-slice@sha256:0102dbe48263fd4e6a62ab9ed48619d10a91c9b4f201ad0dc5b06071c62b05c1", + "operatorId": "operator-builtin-video-slice", + "operatorSlug": "video_slice", + "requiresGpu": false, + "runtimeHandler": "builtin_video_slicing", + "runtimeSource": "backend/operators/builtin/video_slice/operator.py", + "sourceRoot": "backend/operators/builtin/video_slice" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "opencv-python-headless", + "numpy", + "ffmpeg", + "video-files", + "optional-opencv-dnn-model" + ], + "dockerfile": "operator-images/builtin/video_target_clip/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-target-clip:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-video-target-clip@sha256:0eaf4f0fbd20a6f017f1dd77a7d4660dc5cac1112dbbb32fc8e1a4c0cf6377f1", + "operatorId": "operator-builtin-video-target-clip", + "operatorSlug": "video_target_clip", + "requiresGpu": false, + "runtimeHandler": "builtin_video_target_clip", + "runtimeSource": "backend/operators/builtin/video_target_clip/operator.py", + "sourceRoot": "backend/operators/builtin/video_target_clip" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "vla-inference-result-json", + "vla-sample-action-manifest" + ], + "dockerfile": "operator-images/builtin/vla_model_evaluate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-vla-model-evaluate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-vla-model-evaluate@sha256:d940646dabe0e5b7b950e1f32a0ce36110d9dbc379bd123dbb761b8f2e896464", + "operatorId": "operator-builtin-vla-model-evaluate", + "operatorSlug": "vla_model_evaluate", + "requiresGpu": false, + "runtimeHandler": "builtin_vla_model_evaluate", + "runtimeSource": "backend/operators/builtin/vla_model_evaluate/operator.py", + "sourceRoot": "backend/operators/builtin/vla_model_evaluate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "vla-policy-model-json", + "vla-sample-action-manifest" + ], + "dockerfile": "operator-images/builtin/vla_model_inference/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-vla-model-inference:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-vla-model-inference@sha256:6e63bbeca4792ae37589211f3244c0545cbe2217e6f6bcf709b0575275d7d209", + "operatorId": "operator-builtin-vla-model-inference", + "operatorSlug": "vla_model_inference", + "requiresGpu": false, + "runtimeHandler": "builtin_vla_model_inference", + "runtimeSource": "backend/operators/builtin/vla_model_inference/operator.py", + "sourceRoot": "backend/operators/builtin/vla_model_inference" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "vla-sample-action-manifest", + "json-jsonl-csv-files" + ], + "dockerfile": "operator-images/builtin/vla_model_train/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-vla-model-train:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-vla-model-train@sha256:eccc4d24b9a8f0a69d9de173f3eb7b4541550245138babdd68571a77cd91fc45", + "operatorId": "operator-builtin-vla-model-train", + "operatorSlug": "vla_model_train", + "requiresGpu": false, + "runtimeHandler": "builtin_vla_model_train", + "runtimeSource": "backend/operators/builtin/vla_model_train/operator.py", + "sourceRoot": "backend/operators/builtin/vla_model_train" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "json-jsonl-csv-parquet-files", + "image-files", + "pyarrow" + ], + "dockerfile": "operator-images/builtin/vla_sample_manifest_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-vla-sample-manifest-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-vla-sample-manifest-validate@sha256:bcfb95b9a40b9152f2d102c9a00b18e157281e6d7438c84cd7ec6dfb6160df69", + "operatorId": "operator-builtin-vla-sample-manifest-validate", + "operatorSlug": "vla_sample_manifest_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_vla_sample_manifest_validate", + "runtimeSource": "backend/operators/builtin/vla_sample_manifest_validate/operator.py", + "sourceRoot": "backend/operators/builtin/vla_sample_manifest_validate" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "webdataset-tar-shards", + "python-tarfile" + ], + "dockerfile": "operator-images/builtin/webdataset_shard_summary/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-webdataset-shard-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-webdataset-shard-summary@sha256:ce71042733709b064da7e6ceeadfa4a3f4c138d627da5ff9fe7b877d1de2b375", + "operatorId": "operator-builtin-webdataset-shard-summary", + "operatorSlug": "webdataset_shard_summary", + "requiresGpu": false, + "runtimeHandler": "builtin_webdataset_shard_summary", + "runtimeSource": "backend/operators/builtin/webdataset_shard_summary/operator.py", + "sourceRoot": "backend/operators/builtin/webdataset_shard_summary" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "xplatform-dataset-files" + ], + "dockerfile": "operator-images/builtin/xplatform_dataset_index/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-xplatform-dataset-index:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-xplatform-dataset-index@sha256:3dacff68fa10e013999fc940605d2f318d6422e941be392bea05ada8ca3d7908", + "operatorId": "operator-builtin-xplatform-dataset-index", + "operatorSlug": "xplatform_dataset_index", + "requiresGpu": false, + "runtimeHandler": "builtin_xplatform_index", + "runtimeSource": "backend/operators/builtin/xplatform_dataset_index/operator.py", + "sourceRoot": "backend/operators/builtin/xplatform_dataset_index" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "xplatform-dataset-index" + ], + "dockerfile": "operator-images/builtin/xplatform_dataset_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-xplatform-dataset-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-xplatform-dataset-validate@sha256:10a3a8c6a775768205986dd9d81e938c5669be3ada995d03910cc28a3c58aaf3", + "operatorId": "operator-builtin-xplatform-dataset-validate", + "operatorSlug": "xplatform_dataset_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_collection_validate", + "runtimeSource": "backend/operators/builtin/xplatform_dataset_validate/operator.py", + "sourceRoot": "backend/operators/builtin/xplatform_dataset_validate" + }, + { + "baseImage": "ultralytics/ultralytics:latest", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "yolo-label-files", + "image-files", + "pyyaml" + ], + "dockerfile": "operator-images/builtin/yolo_detection_validate/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-yolo-detection-validate:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-yolo-detection-validate@sha256:2478bdf9dd3df22a9d5d6306c8ad0180891efc3d82bd028b50e5abfbc0e7783d", + "operatorId": "operator-builtin-yolo-detection-validate", + "operatorSlug": "yolo_detection_validate", + "requiresGpu": false, + "runtimeHandler": "builtin_yolo_detection_validate", + "runtimeSource": "backend/operators/builtin/yolo_detection_validate/operator.py", + "sourceRoot": "backend/operators/builtin/yolo_detection_validate" + }, + { + "baseImage": "ultralytics/ultralytics:latest", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "real-yolo-training-command", + "yolo-or-coco-dataset", + "gpu-a800" + ], + "dockerfile": "operator-images/builtin/yolo_embodied_detection_train/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-yolo-embodied-detection-train:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-yolo-embodied-detection-train@sha256:52d3ddf838e032dc569a35cf1c937bd34241327a33e6f62efe91fadd1ea9a71a", + "operatorId": "operator-builtin-yolo-embodied-detection-train", + "operatorSlug": "yolo_embodied_detection_train", + "requiresGpu": true, + "runtimeHandler": "builtin_yolo_embodied_detection_train", + "runtimeSource": "backend/operators/builtin/yolo_embodied_detection_train/operator.py", + "sourceRoot": "backend/operators/builtin/yolo_embodied_detection_train" + }, + { + "baseImage": "ultralytics/ultralytics:latest", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "yolo-onnx-model", + "opencv-python-headless", + "numpy", + "ffmpeg", + "video-files", + "gpu-a800" + ], + "dockerfile": "operator-images/builtin/yolo_video_target_clip/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-yolo-video-target-clip:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-yolo-video-target-clip@sha256:36206dcc0ad5f80205273a0e983a66c4c47b65f2cecc94dba64d060016dafa67", + "operatorId": "operator-builtin-yolo-video-target-clip", + "operatorSlug": "yolo_video_target_clip", + "requiresGpu": true, + "runtimeHandler": "builtin_yolo_video_target_clip", + "runtimeSource": "backend/operators/builtin/yolo_video_target_clip/operator.py", + "sourceRoot": "backend/operators/builtin/yolo_video_target_clip" + }, + { + "baseImage": "python:3.11-slim", + "containerArgs": [], + "containerCommand": [ + "python", + "/app/backend/operator-runner/main.py" + ], + "dependencyProfile": [ + "zarr-v2-v3-metadata", + "filesystem" + ], + "dockerfile": "operator-images/builtin/zarr_trajectory_summary/Dockerfile", + "image": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-zarr-trajectory-summary:builtin", + "imageId": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc/droboticflow-operator-zarr-trajectory-summary@sha256:10984c9423a77f6391516f96ad9dd6c708a1d7b0db80a87d5277165513fee505", + "operatorId": "operator-builtin-zarr-trajectory-summary", + "operatorSlug": "zarr_trajectory_summary", + "requiresGpu": false, + "runtimeHandler": "builtin_zarr_trajectory_summary", + "runtimeSource": "backend/operators/builtin/zarr_trajectory_summary/operator.py", + "sourceRoot": "backend/operators/builtin/zarr_trajectory_summary" + } + ], + "registry": "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc", + "schemaVersion": "droboticflow.operator-images/v1", + "tag": "builtin" +} diff --git a/operator-images/builtin/package/Dockerfile b/operator-images/builtin/package/Dockerfile new file mode 100644 index 00000000..122f1a8c --- /dev/null +++ b/operator-images/builtin/package/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-package \ + DROBOTICFLOW_OPERATOR_SLUG=package \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_result_package \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/parquet_table_summary/Dockerfile b/operator-images/builtin/parquet_table_summary/Dockerfile new file mode 100644 index 00000000..bfaadf3c --- /dev/null +++ b/operator-images/builtin/parquet_table_summary/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-parquet-table-summary \ + DROBOTICFLOW_OPERATOR_SLUG=parquet_table_summary \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_parquet_table_summary \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/partnet_mobility_asset_validate/Dockerfile b/operator-images/builtin/partnet_mobility_asset_validate/Dockerfile new file mode 100644 index 00000000..a47bea66 --- /dev/null +++ b/operator-images/builtin/partnet_mobility_asset_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-partnet-mobility-asset-validate \ + DROBOTICFLOW_OPERATOR_SLUG=partnet_mobility_asset_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_partnet_mobility_asset_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/pointcloud_summary/Dockerfile b/operator-images/builtin/pointcloud_summary/Dockerfile new file mode 100644 index 00000000..e3e6ba8f --- /dev/null +++ b/operator-images/builtin/pointcloud_summary/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-pointcloud-summary \ + DROBOTICFLOW_OPERATOR_SLUG=pointcloud_summary \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_pointcloud_summary \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/real_source/Dockerfile b/operator-images/builtin/real_source/Dockerfile new file mode 100644 index 00000000..fa7f85ac --- /dev/null +++ b/operator-images/builtin/real_source/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-real-source \ + DROBOTICFLOW_OPERATOR_SLUG=real_source \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_source \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/rgbd_frame_pair_validate/Dockerfile b/operator-images/builtin/rgbd_frame_pair_validate/Dockerfile new file mode 100644 index 00000000..0e867458 --- /dev/null +++ b/operator-images/builtin/rgbd_frame_pair_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-rgbd-frame-pair-validate \ + DROBOTICFLOW_OPERATOR_SLUG=rgbd_frame_pair_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_rgbd_frame_pair_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/rh20t_scene_validate/Dockerfile b/operator-images/builtin/rh20t_scene_validate/Dockerfile new file mode 100644 index 00000000..454f3f3d --- /dev/null +++ b/operator-images/builtin/rh20t_scene_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-rh20t-scene-validate \ + DROBOTICFLOW_OPERATOR_SLUG=rh20t_scene_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_rh20t_scene_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/rlbench_dataset_validate/Dockerfile b/operator-images/builtin/rlbench_dataset_validate/Dockerfile new file mode 100644 index 00000000..583b9f75 --- /dev/null +++ b/operator-images/builtin/rlbench_dataset_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-rlbench-dataset-validate \ + DROBOTICFLOW_OPERATOR_SLUG=rlbench_dataset_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_rlbench_dataset_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/rlds_episode_summary/Dockerfile b/operator-images/builtin/rlds_episode_summary/Dockerfile new file mode 100644 index 00000000..9925b96a --- /dev/null +++ b/operator-images/builtin/rlds_episode_summary/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-rlds-episode-summary \ + DROBOTICFLOW_OPERATOR_SLUG=rlds_episode_summary \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_rlds_episode_summary \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/robocasa_hdf5_demonstration_validate/Dockerfile b/operator-images/builtin/robocasa_hdf5_demonstration_validate/Dockerfile new file mode 100644 index 00000000..080b295d --- /dev/null +++ b/operator-images/builtin/robocasa_hdf5_demonstration_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-robocasa-hdf5-demonstration-validate \ + DROBOTICFLOW_OPERATOR_SLUG=robocasa_hdf5_demonstration_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_robocasa_hdf5_demonstration_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/robomimic_hdf5_validate/Dockerfile b/operator-images/builtin/robomimic_hdf5_validate/Dockerfile new file mode 100644 index 00000000..0ba084ba --- /dev/null +++ b/operator-images/builtin/robomimic_hdf5_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-robomimic-hdf5-validate \ + DROBOTICFLOW_OPERATOR_SLUG=robomimic_hdf5_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_robomimic_hdf5_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/robomind_hdf5_quality_validate/Dockerfile b/operator-images/builtin/robomind_hdf5_quality_validate/Dockerfile new file mode 100644 index 00000000..d67e0c4a --- /dev/null +++ b/operator-images/builtin/robomind_hdf5_quality_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-robomind-hdf5-quality-validate \ + DROBOTICFLOW_OPERATOR_SLUG=robomind_hdf5_quality_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_robomind_hdf5_quality_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/roboset_hdf5_validate/Dockerfile b/operator-images/builtin/roboset_hdf5_validate/Dockerfile new file mode 100644 index 00000000..944eb624 --- /dev/null +++ b/operator-images/builtin/roboset_hdf5_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-roboset-hdf5-validate \ + DROBOTICFLOW_OPERATOR_SLUG=roboset_hdf5_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_roboset_hdf5_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/robot_model_asset_validate/Dockerfile b/operator-images/builtin/robot_model_asset_validate/Dockerfile new file mode 100644 index 00000000..0b0e7898 --- /dev/null +++ b/operator-images/builtin/robot_model_asset_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-robot-model-asset-validate \ + DROBOTICFLOW_OPERATOR_SLUG=robot_model_asset_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_robot_model_asset_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/roboturk_real_hdf5_validate/Dockerfile b/operator-images/builtin/roboturk_real_hdf5_validate/Dockerfile new file mode 100644 index 00000000..776aee7b --- /dev/null +++ b/operator-images/builtin/roboturk_real_hdf5_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-roboturk-real-hdf5-validate \ + DROBOTICFLOW_OPERATOR_SLUG=roboturk_real_hdf5_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_roboturk_real_hdf5_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/ros_joint_state_validate/Dockerfile b/operator-images/builtin/ros_joint_state_validate/Dockerfile new file mode 100644 index 00000000..17c5ce69 --- /dev/null +++ b/operator-images/builtin/ros_joint_state_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-ros-joint-state-validate \ + DROBOTICFLOW_OPERATOR_SLUG=ros_joint_state_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_ros_joint_state_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/rosbag_summary/Dockerfile b/operator-images/builtin/rosbag_summary/Dockerfile new file mode 100644 index 00000000..d16e4e89 --- /dev/null +++ b/operator-images/builtin/rosbag_summary/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-rosbag-summary \ + DROBOTICFLOW_OPERATOR_SLUG=rosbag_summary \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_rosbag_summary \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/rtx_action_schema_validate/Dockerfile b/operator-images/builtin/rtx_action_schema_validate/Dockerfile new file mode 100644 index 00000000..d9ed26aa --- /dev/null +++ b/operator-images/builtin/rtx_action_schema_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-rtx-action-schema-validate \ + DROBOTICFLOW_OPERATOR_SLUG=rtx_action_schema_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_rtx_action_schema_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/segmentation_mask_validate/Dockerfile b/operator-images/builtin/segmentation_mask_validate/Dockerfile new file mode 100644 index 00000000..843ecc07 --- /dev/null +++ b/operator-images/builtin/segmentation_mask_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-segmentation-mask-validate \ + DROBOTICFLOW_OPERATOR_SLUG=segmentation_mask_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_segmentation_mask_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/sensor_timestamp_sync_validate/Dockerfile b/operator-images/builtin/sensor_timestamp_sync_validate/Dockerfile new file mode 100644 index 00000000..17892d64 --- /dev/null +++ b/operator-images/builtin/sensor_timestamp_sync_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-sensor-timestamp-sync-validate \ + DROBOTICFLOW_OPERATOR_SLUG=sensor_timestamp_sync_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_sensor_timestamp_sync_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/shell_command/Dockerfile b/operator-images/builtin/shell_command/Dockerfile new file mode 100644 index 00000000..88f82c8c --- /dev/null +++ b/operator-images/builtin/shell_command/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-shell-command \ + DROBOTICFLOW_OPERATOR_SLUG=shell_command \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_shell_command \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/sim3d_asset_generate/Dockerfile b/operator-images/builtin/sim3d_asset_generate/Dockerfile new file mode 100644 index 00000000..3330c020 --- /dev/null +++ b/operator-images/builtin/sim3d_asset_generate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-sim3d-asset-generate \ + DROBOTICFLOW_OPERATOR_SLUG=sim3d_asset_generate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_sim3d_asset_generate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/template_render/Dockerfile b/operator-images/builtin/template_render/Dockerfile new file mode 100644 index 00000000..85a47c7c --- /dev/null +++ b/operator-images/builtin/template_render/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-template-render \ + DROBOTICFLOW_OPERATOR_SLUG=template_render \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_text_template \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/tf_transform_tree_validate/Dockerfile b/operator-images/builtin/tf_transform_tree_validate/Dockerfile new file mode 100644 index 00000000..d04a2d7c --- /dev/null +++ b/operator-images/builtin/tf_transform_tree_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-tf-transform-tree-validate \ + DROBOTICFLOW_OPERATOR_SLUG=tf_transform_tree_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_tf_transform_tree_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/tfrecord_structural_validate/Dockerfile b/operator-images/builtin/tfrecord_structural_validate/Dockerfile new file mode 100644 index 00000000..17f33e1b --- /dev/null +++ b/operator-images/builtin/tfrecord_structural_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-tfrecord-structural-validate \ + DROBOTICFLOW_OPERATOR_SLUG=tfrecord_structural_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_tfrecord_structural_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/trajectory_integrity_validate/Dockerfile b/operator-images/builtin/trajectory_integrity_validate/Dockerfile new file mode 100644 index 00000000..80193abd --- /dev/null +++ b/operator-images/builtin/trajectory_integrity_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-trajectory-integrity-validate \ + DROBOTICFLOW_OPERATOR_SLUG=trajectory_integrity_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_trajectory_integrity_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/umi_zarr_dataset_validate/Dockerfile b/operator-images/builtin/umi_zarr_dataset_validate/Dockerfile new file mode 100644 index 00000000..7687afde --- /dev/null +++ b/operator-images/builtin/umi_zarr_dataset_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-umi-zarr-dataset-validate \ + DROBOTICFLOW_OPERATOR_SLUG=umi_zarr_dataset_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_umi_zarr_dataset_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/upload/Dockerfile b/operator-images/builtin/upload/Dockerfile new file mode 100644 index 00000000..55a55133 --- /dev/null +++ b/operator-images/builtin/upload/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-upload \ + DROBOTICFLOW_OPERATOR_SLUG=upload \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_upload \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/video_annotate/Dockerfile b/operator-images/builtin/video_annotate/Dockerfile new file mode 100644 index 00000000..b6392a77 --- /dev/null +++ b/operator-images/builtin/video_annotate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-video-annotate \ + DROBOTICFLOW_OPERATOR_SLUG=video_annotate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_video_annotation \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/video_filter/Dockerfile b/operator-images/builtin/video_filter/Dockerfile new file mode 100644 index 00000000..469a26d7 --- /dev/null +++ b/operator-images/builtin/video_filter/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-video-filter \ + DROBOTICFLOW_OPERATOR_SLUG=video_filter \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_dataset_filter \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/video_quality/Dockerfile b/operator-images/builtin/video_quality/Dockerfile new file mode 100644 index 00000000..844dcdab --- /dev/null +++ b/operator-images/builtin/video_quality/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-video-quality \ + DROBOTICFLOW_OPERATOR_SLUG=video_quality \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_video_quality \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/video_slice/Dockerfile b/operator-images/builtin/video_slice/Dockerfile new file mode 100644 index 00000000..62113623 --- /dev/null +++ b/operator-images/builtin/video_slice/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-video-slice \ + DROBOTICFLOW_OPERATOR_SLUG=video_slice \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_video_slicing \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/video_target_clip/Dockerfile b/operator-images/builtin/video_target_clip/Dockerfile new file mode 100644 index 00000000..6790646f --- /dev/null +++ b/operator-images/builtin/video_target_clip/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-video-target-clip \ + DROBOTICFLOW_OPERATOR_SLUG=video_target_clip \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_video_target_clip \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/vla_model_evaluate/Dockerfile b/operator-images/builtin/vla_model_evaluate/Dockerfile new file mode 100644 index 00000000..5a6b03ab --- /dev/null +++ b/operator-images/builtin/vla_model_evaluate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-vla-model-evaluate \ + DROBOTICFLOW_OPERATOR_SLUG=vla_model_evaluate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_vla_model_evaluate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/vla_model_inference/Dockerfile b/operator-images/builtin/vla_model_inference/Dockerfile new file mode 100644 index 00000000..18fa80d6 --- /dev/null +++ b/operator-images/builtin/vla_model_inference/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-vla-model-inference \ + DROBOTICFLOW_OPERATOR_SLUG=vla_model_inference \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_vla_model_inference \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/vla_model_train/Dockerfile b/operator-images/builtin/vla_model_train/Dockerfile new file mode 100644 index 00000000..ab71c5e6 --- /dev/null +++ b/operator-images/builtin/vla_model_train/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-vla-model-train \ + DROBOTICFLOW_OPERATOR_SLUG=vla_model_train \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_vla_model_train \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/vla_sample_manifest_validate/Dockerfile b/operator-images/builtin/vla_sample_manifest_validate/Dockerfile new file mode 100644 index 00000000..b9432686 --- /dev/null +++ b/operator-images/builtin/vla_sample_manifest_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-vla-sample-manifest-validate \ + DROBOTICFLOW_OPERATOR_SLUG=vla_sample_manifest_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_vla_sample_manifest_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/webdataset_shard_summary/Dockerfile b/operator-images/builtin/webdataset_shard_summary/Dockerfile new file mode 100644 index 00000000..aa2110a8 --- /dev/null +++ b/operator-images/builtin/webdataset_shard_summary/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-webdataset-shard-summary \ + DROBOTICFLOW_OPERATOR_SLUG=webdataset_shard_summary \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_webdataset_shard_summary \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/xplatform_dataset_index/Dockerfile b/operator-images/builtin/xplatform_dataset_index/Dockerfile new file mode 100644 index 00000000..cb0bae74 --- /dev/null +++ b/operator-images/builtin/xplatform_dataset_index/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-xplatform-dataset-index \ + DROBOTICFLOW_OPERATOR_SLUG=xplatform_dataset_index \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_xplatform_index \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/xplatform_dataset_validate/Dockerfile b/operator-images/builtin/xplatform_dataset_validate/Dockerfile new file mode 100644 index 00000000..4b25f220 --- /dev/null +++ b/operator-images/builtin/xplatform_dataset_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-xplatform-dataset-validate \ + DROBOTICFLOW_OPERATOR_SLUG=xplatform_dataset_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_collection_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/yolo_detection_validate/Dockerfile b/operator-images/builtin/yolo_detection_validate/Dockerfile new file mode 100644 index 00000000..798e2905 --- /dev/null +++ b/operator-images/builtin/yolo_detection_validate/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM ultralytics/ultralytics:latest + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-yolo-detection-validate \ + DROBOTICFLOW_OPERATOR_SLUG=yolo_detection_validate \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_yolo_detection_validate \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py ultralytics +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/yolo_embodied_detection_train/Dockerfile b/operator-images/builtin/yolo_embodied_detection_train/Dockerfile new file mode 100644 index 00000000..90142a8a --- /dev/null +++ b/operator-images/builtin/yolo_embodied_detection_train/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM ultralytics/ultralytics:latest + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-yolo-embodied-detection-train \ + DROBOTICFLOW_OPERATOR_SLUG=yolo_embodied_detection_train \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_yolo_embodied_detection_train \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py ultralytics +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/yolo_video_target_clip/Dockerfile b/operator-images/builtin/yolo_video_target_clip/Dockerfile new file mode 100644 index 00000000..4b189722 --- /dev/null +++ b/operator-images/builtin/yolo_video_target_clip/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM ultralytics/ultralytics:latest + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-yolo-video-target-clip \ + DROBOTICFLOW_OPERATOR_SLUG=yolo_video_target_clip \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_yolo_video_target_clip \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py ultralytics +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/operator-images/builtin/zarr_trajectory_summary/Dockerfile b/operator-images/builtin/zarr_trajectory_summary/Dockerfile new file mode 100644 index 00000000..fbe3464d --- /dev/null +++ b/operator-images/builtin/zarr_trajectory_summary/Dockerfile @@ -0,0 +1,19 @@ +# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM python:3.11-slim + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \ + PYTHONPATH=/app \ + DROBOTICFLOW_OPERATOR_ID=operator-builtin-zarr-trajectory-summary \ + DROBOTICFLOW_OPERATOR_SLUG=zarr_trajectory_summary \ + DROBOTICFLOW_OPERATOR_HANDLER=builtin_zarr_trajectory_summary \ + USER=droboticflow \ + LOGNAME=droboticflow \ + HOME=/tmp/droboticflow-home + +RUN apt-get update && apt-get install -y --no-install-recommends ca-certificates ffmpeg && rm -rf /var/lib/apt/lists/* +RUN python -m pip install --no-cache-dir --upgrade pip \ + && python -m pip install --no-cache-dir numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] diff --git a/package.json b/package.json index 95a954f1..662471ce 100644 --- a/package.json +++ b/package.json @@ -29,6 +29,8 @@ "test:frontend": "node --test tests/frontend/*.test.mjs", "operators:manifests": "python3 scripts/generate_operator_repository_manifests.py --write", "operators:manifests:check": "python3 scripts/generate_operator_repository_manifests.py --check", + "operators:images": "python3 scripts/generate_operator_images.py --write", + "operators:images:check": "python3 scripts/generate_operator_images.py --check", "audit:moderate": "./scripts/npm-audit-retry.sh --audit-level=moderate", "verify:local": "npm run lint && npm run format:check && npm test && npm run test:python && npm run test:frontend && npm run build && npm run k8s:validate", "verify:security": "npm run audit:moderate", diff --git a/scripts/build-poc-images-kaniko.sh b/scripts/build-poc-images-kaniko.sh index 0e647694..fc90ea67 100755 --- a/scripts/build-poc-images-kaniko.sh +++ b/scripts/build-poc-images-kaniko.sh @@ -25,6 +25,9 @@ POC_KANIKO_SKIP_DEFAULT_REGISTRY_FALLBACK="${POC_KANIKO_SKIP_DEFAULT_REGISTRY_FA POC_KANIKO_CLEAN_CONTEXT="${POC_KANIKO_CLEAN_CONTEXT:-true}" POC_KANIKO_WAIT_TIMEOUT="${POC_KANIKO_WAIT_TIMEOUT:-1200s}" POC_KANIKO_GOPROXY="${POC_KANIKO_GOPROXY:-}" +POC_KANIKO_CONTEXT_SOURCE="${POC_KANIKO_CONTEXT_SOURCE:-worktree}" +POC_KANIKO_BUILD_OPERATORS="${POC_KANIKO_BUILD_OPERATORS:-false}" +POC_KANIKO_OPERATOR_MANIFEST="${POC_KANIKO_OPERATOR_MANIFEST:-operator-images/builtin/operator-images.json}" POC_BUILD_CONTEXT_ROOT="${POC_BUILD_CONTEXT_ROOT:-$WORKSPACE_MOUNT_PATH/build-contexts}" if [[ -z "${IMAGE_TAG:-}" ]]; then @@ -47,6 +50,15 @@ validate_image_reference WORKER_IMAGE "$WORKER_IMAGE" validate_image_reference OPERATOR_RUNNER_IMAGE "$OPERATOR_RUNNER_IMAGE" validate_boolean POC_KANIKO_SKIP_DEFAULT_REGISTRY_FALLBACK "$POC_KANIKO_SKIP_DEFAULT_REGISTRY_FALLBACK" validate_boolean POC_KANIKO_CLEAN_CONTEXT "$POC_KANIKO_CLEAN_CONTEXT" +validate_boolean POC_KANIKO_BUILD_OPERATORS "$POC_KANIKO_BUILD_OPERATORS" +case "$POC_KANIKO_CONTEXT_SOURCE" in + worktree|head) + ;; + *) + echo "ERROR: POC_KANIKO_CONTEXT_SOURCE must be worktree or head." >&2 + exit 1 + ;; +esac if [[ "$POC_KANIKO_GOPROXY" =~ [[:space:]] ]]; then echo "ERROR: POC_KANIKO_GOPROXY must not contain whitespace." >&2 exit 1 @@ -57,6 +69,11 @@ if [[ "$POC_BUILD_CONTEXT_ROOT" != "$WORKSPACE_MOUNT_PATH" && "$POC_BUILD_CONTEX exit 1 fi +if boolean_true POC_KANIKO_BUILD_OPERATORS "$POC_KANIKO_BUILD_OPERATORS" && [[ ! -f "$ROOT_DIR/$POC_KANIKO_OPERATOR_MANIFEST" ]]; then + echo "ERROR: POC_KANIKO_OPERATOR_MANIFEST does not exist: $POC_KANIKO_OPERATOR_MANIFEST" >&2 + exit 1 +fi + CONTEXT_DIR="${POC_BUILD_CONTEXT_ROOT%/}/droboticflow-${IMAGE_TAG}" api_pod="" context_json="" @@ -201,6 +218,43 @@ wait_for_job() { done } +operator_image_specs() { + python3 - "$ROOT_DIR/$POC_KANIKO_OPERATOR_MANIFEST" "$REGISTRY_HOST" "$REGISTRY_NAMESPACE" "$IMAGE_TAG" <<'PY' +import json +import pathlib +import sys + +manifest_path = pathlib.Path(sys.argv[1]) +registry_host = sys.argv[2].strip() +registry_namespace = sys.argv[3].strip() +tag = sys.argv[4].strip() +payload = json.loads(manifest_path.read_text(encoding="utf-8")) +if not isinstance(payload, dict) or not isinstance(payload.get("items"), list): + raise SystemExit(f"{manifest_path} must contain items[]") +if not registry_host or not registry_namespace: + raise SystemExit("REGISTRY_HOST and REGISTRY_NAMESPACE are required for operator image builds") +for item in sorted(payload["items"], key=lambda value: value.get("operatorSlug", "")): + slug = str(item.get("operatorSlug", "")).strip() + dockerfile = str(item.get("dockerfile", "")).strip() + if not slug or not dockerfile: + raise SystemExit(f"invalid operator image manifest item: {item!r}") + image = f"{registry_host}/{registry_namespace}/droboticflow-operator-{slug.replace('_', '-')}:{tag}" + component = f"op-{slug.replace('_', '-')}" + print("\t".join([component, dockerfile, image])) +PY +} + +upload_build_context() { + case "$POC_KANIKO_CONTEXT_SOURCE" in + head) + git -C "$ROOT_DIR" archive --format=tar HEAD + ;; + worktree) + git -C "$ROOT_DIR" ls-files -z | ROOT_DIR="$ROOT_DIR" python3 -c 'import os,pathlib,sys; root=pathlib.Path(os.environ["ROOT_DIR"]); [sys.stdout.buffer.write(item + b"\0") for item in sys.stdin.buffer.read().split(b"\0") if item and (root / item.decode("utf-8", errors="surrogateescape")).exists()]' | COPYFILE_DISABLE=1 tar -C "$ROOT_DIR" --null -T - -cf - + ;; + esac +} + cleanup_context() { if [[ "${context_uploaded:-false}" != "true" ]]; then return 0 @@ -224,17 +278,41 @@ fi context_json="$(json_string "$CONTEXT_DIR")" kubectl -n "$K8S_NAMESPACE" exec "$api_pod" -- sh -lc "rm -rf $context_json && mkdir -p $context_json" context_uploaded=true -git -C "$ROOT_DIR" archive --format=tar HEAD | kubectl -n "$K8S_NAMESPACE" exec -i "$api_pod" -- tar -xf - -C "$CONTEXT_DIR" +upload_build_context | kubectl -n "$K8S_NAMESPACE" exec -i "$api_pod" -- tar -xf - -C "$CONTEXT_DIR" create_kaniko_job api Dockerfile.api "$API_IMAGE" create_kaniko_job worker Dockerfile.worker "$WORKER_IMAGE" create_kaniko_job runner Dockerfile.operator-runner "$OPERATOR_RUNNER_IMAGE" +operator_components=() +operator_component_count=0 +if boolean_true POC_KANIKO_BUILD_OPERATORS "$POC_KANIKO_BUILD_OPERATORS"; then + while IFS=$'\t' read -r component dockerfile image; do + if [[ -z "$component" ]]; then + continue + fi + validate_image_reference "operator image $component" "$image" + create_kaniko_job "$component" "$dockerfile" "$image" + operator_components+=("$component:$image") + operator_component_count=$((operator_component_count + 1)) + done < <(operator_image_specs) +fi + wait_for_job api wait_for_job worker wait_for_job runner +if (( operator_component_count > 0 )); then + for spec in "${operator_components[@]}"; do + wait_for_job "${spec%%:*}" + done +fi printf 'IMAGE_TAG=%s\n' "$IMAGE_TAG" printf 'API_IMAGE=%s\n' "$API_IMAGE" printf 'WORKER_IMAGE=%s\n' "$WORKER_IMAGE" printf 'OPERATOR_RUNNER_IMAGE=%s\n' "$OPERATOR_RUNNER_IMAGE" +if (( operator_component_count > 0 )); then + for spec in "${operator_components[@]}"; do + printf 'OPERATOR_IMAGE=%s\n' "${spec#*:}" + done +fi diff --git a/scripts/generate_operator_images.py b/scripts/generate_operator_images.py new file mode 100644 index 00000000..4ec90f18 --- /dev/null +++ b/scripts/generate_operator_images.py @@ -0,0 +1,339 @@ +#!/usr/bin/env python3 +from __future__ import annotations + +import argparse +import hashlib +import importlib.util +import json +import pathlib +import sys +from typing import Any + + +ROOT = pathlib.Path(__file__).resolve().parents[1] +CATALOG_PATH = ROOT / "internal" / "api" / "builtin_operators_catalog.json" +OPERATOR_ROOT = ROOT / "backend" / "operators" / "builtin" +IMAGE_ROOT = ROOT / "operator-images" / "builtin" +MANIFEST_PATH = IMAGE_ROOT / "operator-images.json" +DEFAULT_REGISTRY = "ccr-29eug8s3-pub.cnc.bj.baidubce.com/dataflow-poc" +IMAGE_TAG = "builtin" +COMMAND_ENTRYPOINT = ["python", "/app/backend/operator-runner/main.py"] +GPU_SLUGS = { + "act_policy_train", + "diffusion_policy_train", + "openvla_finetune", + "openvla_inference", + "yolo_embodied_detection_train", + "yolo_video_target_clip", +} +COMMAND_WRAPPER_ALLOWED = {"operator-builtin-shell-command"} +STRUCTURED_OPERATOR_OVERRIDES: dict[str, dict[str, Any]] = { + "operator-builtin-model-inference": { + "skillDefinition": "执行结构化模型推理节点;读取 MODEL_JSON 或 MODEL_PATH,支持 rule、linear、nearest_centroid 三类可解释模型;自定义命令请使用 Shell 命令算子,constant 等硬编码输出模型会被拒绝。", + }, + "operator-builtin-act-policy-train": { + "skillDefinition": "ACT 策略训练算子;使用 DATASET_PATH/STATE_FIELD/ACTION_FIELD 等结构化参数读取真实轨迹样本,在独立 GPU 镜像中训练并输出 checkpoint.pt 与 metrics.json。推荐 1 张 A800 80GB。", + "hyperparams": [ + {"key": "DATASET_PATH", "value": "", "description": "真实训练数据路径;支持 JSON/JSONL/CSV 或包含 samples.json/jsonl 的目录", "type": "string"}, + {"key": "STATE_FIELD", "value": "state", "description": "状态向量字段名", "type": "string"}, + {"key": "ACTION_FIELD", "value": "action", "description": "动作向量字段名", "type": "string"}, + {"key": "EPOCHS", "value": "5", "description": "训练轮数", "type": "number"}, + {"key": "BATCH_SIZE", "value": "8", "description": "训练 batch size", "type": "number"}, + {"key": "LEARNING_RATE", "value": "0.001", "description": "AdamW 学习率", "type": "number"}, + {"key": "HIDDEN_DIM", "value": "128", "description": "策略网络隐藏层宽度", "type": "number"}, + {"key": "CHUNK_SIZE", "value": "1", "description": "ACT action chunk size", "type": "number"}, + {"key": "RECOMMENDED_GPU", "value": "1x NVIDIA A800 80GB", "description": "推荐 GPU 配置", "type": "string"}, + ], + }, + "operator-builtin-diffusion-policy-train": { + "skillDefinition": "Diffusion Policy 训练算子;使用 DATASET_PATH/STATE_FIELD/ACTION_FIELD/HORIZON/DIFFUSION_STEPS 等结构化参数读取真实轨迹数据,输出 checkpoint.pt 与 metrics.json。推荐 1 张 A800 80GB。", + "hyperparams": [ + {"key": "DATASET_PATH", "value": "", "description": "真实训练数据路径;支持 JSON/JSONL/CSV 或包含 samples.json/jsonl 的目录", "type": "string"}, + {"key": "STATE_FIELD", "value": "state", "description": "观测/状态向量字段名", "type": "string"}, + {"key": "ACTION_FIELD", "value": "action", "description": "动作向量字段名", "type": "string"}, + {"key": "EPOCHS", "value": "5", "description": "训练轮数", "type": "number"}, + {"key": "BATCH_SIZE", "value": "8", "description": "训练 batch size", "type": "number"}, + {"key": "LEARNING_RATE", "value": "0.001", "description": "AdamW 学习率", "type": "number"}, + {"key": "HIDDEN_DIM", "value": "256", "description": "策略网络隐藏层宽度", "type": "number"}, + {"key": "HORIZON", "value": "16", "description": "预测 horizon", "type": "number"}, + {"key": "DIFFUSION_STEPS", "value": "32", "description": "扩散步数配置,用于训练记录和下游复现", "type": "number"}, + {"key": "RECOMMENDED_GPU", "value": "1x NVIDIA A800 80GB", "description": "推荐 GPU 配置", "type": "string"}, + ], + }, + "operator-builtin-openvla-finetune": { + "skillDefinition": "OpenVLA 微调算子;使用 BASE_MODEL、DATASET_PATH、TEXT_FIELD、TARGET_FIELD 等结构化参数,通过 Transformers/PyTorch 对真实指令-动作样本训练并输出模型目录与 metrics.json。推荐 1 张 A800 80GB。", + "hyperparams": [ + {"key": "BASE_MODEL", "value": "", "description": "Hugging Face 或本地基础模型路径", "type": "string"}, + {"key": "DATASET_PATH", "value": "", "description": "真实微调数据 JSON/JSONL/CSV 路径", "type": "string"}, + {"key": "TEXT_FIELD", "value": "instruction", "description": "指令/观测文本字段名", "type": "string"}, + {"key": "TARGET_FIELD", "value": "action", "description": "目标动作文本字段名", "type": "string"}, + {"key": "MAX_LENGTH", "value": "256", "description": "token 最大长度", "type": "number"}, + {"key": "EPOCHS", "value": "1", "description": "训练轮数", "type": "number"}, + {"key": "BATCH_SIZE", "value": "1", "description": "训练 batch size", "type": "number"}, + {"key": "LEARNING_RATE", "value": "0.00005", "description": "AdamW 学习率", "type": "number"}, + {"key": "PRECISION", "value": "bf16", "description": "推荐精度配置", "type": "string"}, + {"key": "RECOMMENDED_GPU", "value": "1x NVIDIA A800 80GB", "description": "推荐 GPU 配置", "type": "string"}, + ], + }, + "operator-builtin-openvla-inference": { + "skillDefinition": "OpenVLA 推理算子;使用 MODEL_PATH 或 BASE_MODEL 加载真实 Transformers 权重,读取 PROMPT 或上游样本 instruction,并输出 inference-result.json。推荐 1 张 A800,量化模型可降低显存。", + "hyperparams": [ + {"key": "MODEL_PATH", "value": "", "description": "本地模型目录或 Hugging Face 模型名", "type": "string"}, + {"key": "BASE_MODEL", "value": "", "description": "未提供 MODEL_PATH 时使用的基础模型", "type": "string"}, + {"key": "PROMPT", "value": "", "description": "推理 prompt;为空则从上游样本读取 TEXT_FIELD", "type": "string"}, + {"key": "TEXT_FIELD", "value": "instruction", "description": "上游样本指令字段名", "type": "string"}, + {"key": "MAX_NEW_TOKENS", "value": "64", "description": "生成 token 数", "type": "number"}, + {"key": "PRECISION", "value": "bf16", "description": "推荐精度配置", "type": "string"}, + {"key": "RECOMMENDED_GPU", "value": "1x NVIDIA A800 80GB", "description": "推荐 GPU 配置", "type": "string"}, + ], + }, + "operator-builtin-yolo-embodied-detection-train": { + "skillDefinition": "YOLO 具身目标检测训练算子;使用 Ultralytics API 读取真实 YOLO/COCO data.yaml,输出 best/last checkpoint 与 metrics.json。A800 对 YOLOv8/YOLOv10/YOLO11 中大模型训练充足,默认请求 1 卡。", + "hyperparams": [ + {"key": "DATA_YAML", "value": "", "description": "真实 YOLO/COCO data.yaml 路径", "type": "string"}, + {"key": "MODEL_SIZE", "value": "yolo11n.pt", "description": "Ultralytics 模型名或权重路径", "type": "string"}, + {"key": "EPOCHS", "value": "10", "description": "训练轮数", "type": "number"}, + {"key": "IMAGE_SIZE", "value": "640", "description": "输入图片尺寸", "type": "number"}, + {"key": "BATCH_SIZE", "value": "8", "description": "训练 batch size", "type": "number"}, + {"key": "CONF", "value": "0.25", "description": "后续检测推荐置信度", "type": "number"}, + {"key": "TARGET_LABELS", "value": "", "description": "关注类别,逗号分隔", "type": "string"}, + {"key": "RECOMMENDED_GPU", "value": "1x NVIDIA A800 80GB", "description": "推荐 GPU 配置", "type": "string"}, + ], + }, +} + + +def load_repository_generator() -> Any: + path = ROOT / "scripts" / "generate_operator_repository_manifests.py" + spec = importlib.util.spec_from_file_location("operator_repository_generator", path) + if spec is None or spec.loader is None: + raise RuntimeError(f"cannot load {path}") + module = importlib.util.module_from_spec(spec) + spec.loader.exec_module(module) + return module + + +REPOSITORY_GENERATOR = load_repository_generator() +RUNTIME_DEPENDENCIES: dict[str, list[str]] = REPOSITORY_GENERATOR.RUNTIME_DEPENDENCIES + + +def load_json(path: pathlib.Path) -> Any: + return json.loads(path.read_text(encoding="utf-8")) + + +def dump_json(payload: Any) -> str: + return json.dumps(payload, ensure_ascii=False, indent=2, sort_keys=True) + "\n" + + +def operator_slug(operator_id: str) -> str: + return operator_id.removeprefix("operator-builtin-").replace("-", "_") + + +def image_name(slug: str, registry: str = DEFAULT_REGISTRY) -> str: + return f"{registry}/droboticflow-operator-{slug.replace('_', '-')}:{IMAGE_TAG}" + + +def image_id(slug: str, registry: str = DEFAULT_REGISTRY) -> str: + # Catalog defaults use a stable source-contract digest. POC reconcile replaces it + # with the real registry image digest after Kaniko build. + digest = hashlib.sha256(f"droboticflow:{slug}:{IMAGE_TAG}".encode("utf-8")).hexdigest() + return f"{registry}/droboticflow-operator-{slug.replace('_', '-')}@sha256:{digest}" + + +def docker_base(slug: str) -> str: + if slug.startswith("yolo_"): + return "ultralytics/ultralytics:latest" + if slug in GPU_SLUGS: + return "pytorch/pytorch:2.5.1-cuda12.4-cudnn9-runtime" + return "python:3.11-slim" + + +def dockerfile(slug: str, operator_id: str, handler: str) -> str: + apt_prefix = "apt-get update && apt-get install -y --no-install-recommends" + apt_suffix = "&& rm -rf /var/lib/apt/lists/*" + pip_packages = "numpy opencv-python-headless bce-python-sdk pymongo h5py pyarrow mcap PyYAML zarr nvidia-ml-py" + if slug.startswith("yolo_"): + pip_packages += " ultralytics" + elif slug in {"act_policy_train", "diffusion_policy_train", "openvla_finetune", "openvla_inference"}: + pip_packages += " torch torchvision transformers accelerate safetensors" + return f"""# Generated by scripts/generate_operator_images.py. Do not edit by hand. +FROM {docker_base(slug)} + +WORKDIR /app +ENV PYTHONUNBUFFERED=1 \\ + PYTHONPATH=/app \\ + DROBOTICFLOW_OPERATOR_ID={operator_id} \\ + DROBOTICFLOW_OPERATOR_SLUG={slug} \\ + DROBOTICFLOW_OPERATOR_HANDLER={handler} \\ + USER=droboticflow \\ + LOGNAME=droboticflow \\ + HOME=/tmp/droboticflow-home + +RUN {apt_prefix} ca-certificates ffmpeg {apt_suffix} +RUN python -m pip install --no-cache-dir --upgrade pip \\ + && python -m pip install --no-cache-dir {pip_packages} +RUN mkdir -p /tmp/droboticflow-home /workspace /outputs +COPY backend ./backend +ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"] +""" + + +def contract_for(item: dict[str, Any], slug: str) -> dict[str, Any]: + dependencies = RUNTIME_DEPENDENCIES.get(slug, []) + gpu = slug in GPU_SLUGS + return { + "schemaVersion": "droboticflow.operator.runtime/v1", + "operatorId": item["id"], + "operatorSlug": slug, + "handler": item.get("runtimeHandler") or slug, + "entrypoint": "/app/backend/operator-runner/main.py", + "imagePolicy": { + "independentFinalImage": True, + "allowsSharedBaseLayer": True, + "forbidsSharedOperatorRunnerFinalImage": True, + }, + "runtimeDependencies": dependencies, + "resources": { + "requiresGpu": gpu, + "recommendedGpuType": "NVIDIA A800 80GB" if gpu else "", + "minGpuCount": 1 if gpu else 0, + }, + "parameters": { + "source": "hyperparams", + "structured": True, + "commandWrapperAllowed": item["id"] in COMMAND_WRAPPER_ALLOWED, + }, + "artifacts": { + "requiresManifest": True, + "requiresChecksum": True, + "requiresLineage": True, + }, + "acceptance": { + "requiresRealFixture": True, + "requiresFixtureChecksum": True, + "requiresArtifactChecksum": True, + "requiresWorkflowTransfer": True, + "requiresGpuEvidence": gpu, + }, + } + + +def strip_disallowed_command_params(item: dict[str, Any]) -> dict[str, Any]: + if item["id"] in COMMAND_WRAPPER_ALLOWED: + return item + hyperparams = [ + param + for param in item.get("hyperparams", []) + if str(param.get("key", "")).strip().upper() not in {"COMMAND", "SCRIPT"} + ] + if len(hyperparams) != len(item.get("hyperparams", [])): + item = {**item, "hyperparams": hyperparams} + return item + + +def expected_catalog() -> list[dict[str, Any]]: + items = load_json(CATALOG_PATH) + next_items: list[dict[str, Any]] = [] + for item in items: + if item.get("type") != "builtin": + next_items.append(item) + continue + slug = operator_slug(item["id"]) + handler = item.get("runtimeHandler") or slug + next_item = strip_disallowed_command_params(dict(item)) + overrides = STRUCTURED_OPERATOR_OVERRIDES.get(next_item["id"], {}) + if "skillDefinition" in overrides: + next_item["skillDefinition"] = overrides["skillDefinition"] + if "hyperparams" in overrides: + next_item["hyperparams"] = overrides["hyperparams"] + next_item["image"] = image_name(slug) + next_item["imageId"] = image_id(slug) + next_item["containerCommand"] = COMMAND_ENTRYPOINT + next_item["containerArgs"] = [] + next_item["runtimeContract"] = contract_for(next_item, slug) + next_items.append(next_item) + return next_items + + +def expected_files(catalog_items: list[dict[str, Any]]) -> dict[pathlib.Path, str]: + files: dict[pathlib.Path, str] = {} + manifest_items: list[dict[str, Any]] = [] + for item in catalog_items: + if item.get("type") != "builtin": + continue + slug = operator_slug(item["id"]) + metadata_path = OPERATOR_ROOT / slug / "operator.json" + if not metadata_path.exists(): + raise RuntimeError(f"missing operator metadata for {item['id']}: {metadata_path.relative_to(ROOT)}") + handler = item.get("runtimeHandler") or slug + files[IMAGE_ROOT / slug / "Dockerfile"] = dockerfile(slug, item["id"], handler) + manifest_items.append( + { + "operatorId": item["id"], + "operatorSlug": slug, + "image": item["image"], + "imageId": item["imageId"], + "dockerfile": f"operator-images/builtin/{slug}/Dockerfile", + "sourceRoot": f"backend/operators/builtin/{slug}", + "runtimeSource": item.get("runtimeSource", f"backend/operators/builtin/{slug}/operator.py"), + "runtimeHandler": handler, + "baseImage": docker_base(slug), + "dependencyProfile": RUNTIME_DEPENDENCIES.get(slug, []), + "requiresGpu": slug in GPU_SLUGS, + "containerCommand": COMMAND_ENTRYPOINT, + "containerArgs": [], + } + ) + files[MANIFEST_PATH] = dump_json( + { + "schemaVersion": "droboticflow.operator-images/v1", + "registry": DEFAULT_REGISTRY, + "tag": IMAGE_TAG, + "items": sorted(manifest_items, key=lambda item: item["operatorId"]), + } + ) + return files + + +def write_files(catalog_items: list[dict[str, Any]], files: dict[pathlib.Path, str]) -> None: + CATALOG_PATH.write_text(dump_json(catalog_items), encoding="utf-8") + for path, content in files.items(): + path.parent.mkdir(parents=True, exist_ok=True) + path.write_text(content, encoding="utf-8") + + +def check_files(catalog_items: list[dict[str, Any]], files: dict[pathlib.Path, str]) -> int: + mismatches: list[str] = [] + expected_catalog_text = dump_json(catalog_items) + if CATALOG_PATH.read_text(encoding="utf-8") != expected_catalog_text: + mismatches.append(f"outdated {CATALOG_PATH.relative_to(ROOT)}") + for path, content in files.items(): + if not path.exists(): + mismatches.append(f"missing {path.relative_to(ROOT)}") + continue + if path.read_text(encoding="utf-8") != content: + mismatches.append(f"outdated {path.relative_to(ROOT)}") + if mismatches: + for mismatch in mismatches: + print(mismatch, file=sys.stderr) + return 1 + return 0 + + +def main() -> int: + parser = argparse.ArgumentParser(description="Generate or check independent builtin operator images.") + mode = parser.add_mutually_exclusive_group() + mode.add_argument("--write", action="store_true", help="write catalog image contracts and Dockerfiles") + mode.add_argument("--check", action="store_true", help="check generated catalog image contracts and Dockerfiles") + args = parser.parse_args() + + catalog_items = expected_catalog() + files = expected_files(catalog_items) + if args.check: + return check_files(catalog_items, files) + write_files(catalog_items, files) + return 0 + + +if __name__ == "__main__": + raise SystemExit(main()) diff --git a/scripts/generate_operator_repository_manifests.py b/scripts/generate_operator_repository_manifests.py index 0e87060e..03d8d184 100644 --- a/scripts/generate_operator_repository_manifests.py +++ b/scripts/generate_operator_repository_manifests.py @@ -166,6 +166,11 @@ def expected_manifests() -> dict[pathlib.Path, dict[str, Any]]: "runtimeSource": runtime_source, "runtimeHandler": metadata["runtimeHandler"], "containerEntrypoint": "backend/operator-runner/main.py", + "image": catalog_item.get("image", ""), + "imageId": catalog_item.get("imageId", ""), + "containerCommand": catalog_item.get("containerCommand", []), + "containerArgs": catalog_item.get("containerArgs", []), + "runtimeContract": catalog_item.get("runtimeContract", {}), "pythonImport": f"backend.operators.builtin.{slug}.operator:run", "runtimeDependencies": dependencies, "strictPocAcceptanceExternals": STRICT_EXTERNALS.get(slug, []), diff --git a/scripts/poc-operator-workflow-acceptance.sh b/scripts/poc-operator-workflow-acceptance.sh index b3b37f10..a0746630 100755 --- a/scripts/poc-operator-workflow-acceptance.sh +++ b/scripts/poc-operator-workflow-acceptance.sh @@ -181,6 +181,40 @@ PY )" response="$(api_curl -X POST "$API_BASE_URL/api/acceptance/poc-operator-workflows" -H 'Content-Type: application/json' -d "$payload")" printf '%s' "$response" | python3 -c 'import json, sys; payload=json.load(sys.stdin); print("created workflow batches:", json.dumps({"operatorCount": payload.get("operatorCount"), "runs": [{"batch": item.get("batch"), "runId": item.get("runId"), "operators": len(item.get("operatorIds") or [])} for item in payload.get("runs", [])]}, ensure_ascii=False))' +RESPONSE="$response" python3 - <<'PY' +import json +import os +import sys + +payload = json.loads(os.environ["RESPONSE"]) +summary = payload.get("summary") or {} +print("operator workflow contract:", json.dumps(summary, ensure_ascii=False)) +operator_count = int(summary.get("operatorCount") or payload.get("operatorCount") or 0) +coverage = int(summary.get("independentImageCoverage") or 0) +required = int(summary.get("independentImageRequired") or operator_count) +command_wrappers = int(summary.get("commandWrapperOnlyOperatorCount") or 0) +generated_fixtures = int(summary.get("generatedFixtureCount") or 0) +mock_fixtures = int(summary.get("mockFixtureCount") or 0) +if operator_count <= 0: + raise SystemExit("operator workflow acceptance did not select any operators") +if coverage != required: + raise SystemExit(f"independent image coverage is incomplete: {coverage}/{required}") +if command_wrappers != 0: + raise SystemExit(f"non-shell command-wrapper operators are not allowed: {command_wrappers}") +if generated_fixtures != 0 or mock_fixtures != 0: + raise SystemExit(f"generated/mock fixtures are not allowed: generated={generated_fixtures} mock={mock_fixtures}") +for run in payload.get("runs") or []: + evidence = run.get("imageEvidence") or [] + if len(evidence) != len(run.get("operatorIds") or []): + raise SystemExit(f"run {run.get('runId')} image evidence count does not match operators") + for item in evidence: + image = str(item.get("image") or "") + image_id = str(item.get("imageId") or "") + if "droboticflow-operator-runner" in image or "droboticflow-operator-" not in image: + raise SystemExit(f"operator {item.get('operatorId')} does not use an independent operator image") + if "@sha256:" not in image_id: + raise SystemExit(f"operator {item.get('operatorId')} imageId is not digest-pinned") +PY run_ids="$(printf '%s' "$response" | python3 -c 'import json, sys; [print(item["runId"]) for item in json.load(sys.stdin).get("runs", [])]')" if [[ -z "$run_ids" ]]; then echo "ERROR: no runs were created." >&2 diff --git a/scripts/poc_strict_preflight_result.py b/scripts/poc_strict_preflight_result.py index 30c8e5fd..767eb98b 100644 --- a/scripts/poc_strict_preflight_result.py +++ b/scripts/poc_strict_preflight_result.py @@ -22,6 +22,7 @@ STAGE_HANDLERS = { } REPOSITORY_POLICY_VIOLATION_KEYS = ( "repositoryManifestErrors", + "operatorImageContractErrors", "duplicateOperatorIds", "duplicateOperatorNames", "legacyOperators", @@ -104,10 +105,12 @@ def operator_helper_builtin_keys() -> list[str]: def static_operator_layout() -> dict[str, Any]: operator_root = ROOT / "backend" / "operators" / "builtin" + catalog_path = ROOT / "internal" / "api" / "builtin_operators_catalog.json" metadata_paths = sorted(operator_root.glob("*/operator.json")) sources: list[str] = [] repository_manifests: list[str] = [] repository_manifest_errors: list[str] = [] + operator_image_contract_errors: list[str] = [] legacy: list[str] = [] builtin_helper_calls: list[str] = [] hardcoded_external_api_sources: list[str] = [] @@ -115,6 +118,19 @@ def static_operator_layout() -> dict[str, Any]: duplicate_operator_ids: list[str] = [] seen_operator_names: dict[str, str] = {} duplicate_operator_names: list[str] = [] + catalog_items: dict[str, dict[str, Any]] = {} + try: + loaded_catalog = json.loads(catalog_path.read_text(encoding="utf-8")) + if isinstance(loaded_catalog, list): + catalog_items = { + str(item.get("id", "")).strip(): item + for item in loaded_catalog + if isinstance(item, dict) and item.get("type") == "builtin" + } + else: + operator_image_contract_errors.append("builtin catalog must be a JSON list") + except Exception as error: + operator_image_contract_errors.append(f"builtin catalog is not readable ({type(error).__name__}: {error})") for metadata_path in metadata_paths: slug = metadata_path.parent.name @@ -187,6 +203,36 @@ def static_operator_layout() -> dict[str, Any]: ) if "externalApi" in metadata and repository_manifest.get("externalApi") != metadata.get("externalApi"): repository_manifest_errors.append(f"{slug}: repository.json externalApi must match operator.json") + catalog_item = catalog_items.get(operator_id) + if not catalog_item: + operator_image_contract_errors.append(f"{slug}: missing builtin catalog item") + catalog_item = {} + image = str(catalog_item.get("image") or repository_manifest.get("image") or "").strip() + image_id = str(catalog_item.get("imageId") or repository_manifest.get("imageId") or "").strip() + runtime_contract = catalog_item.get("runtimeContract") or repository_manifest.get("runtimeContract") + if not image: + operator_image_contract_errors.append(f"{slug}: builtin image is required") + if "droboticflow-operator-runner" in image: + operator_image_contract_errors.append(f"{slug}: builtin image must not use shared operator-runner final image") + if "droboticflow-operator-" not in image: + operator_image_contract_errors.append(f"{slug}: builtin image must use an independent operator image tag") + if not re.search(r"@sha256:[a-fA-F0-9]{64}$", image_id): + operator_image_contract_errors.append(f"{slug}: imageId must be digest-pinned") + if catalog_item.get("containerCommand") != ["python", "/app/backend/operator-runner/main.py"]: + operator_image_contract_errors.append(f"{slug}: containerCommand must point at the operator runner entrypoint") + if catalog_item.get("containerArgs", []) != []: + operator_image_contract_errors.append(f"{slug}: containerArgs must default to []") + if not isinstance(runtime_contract, dict): + operator_image_contract_errors.append(f"{slug}: runtimeContract is required") + elif not runtime_contract.get("imagePolicy", {}).get("independentFinalImage"): + operator_image_contract_errors.append(f"{slug}: runtimeContract must require independent final image") + command_keys = { + str(param.get("key", "")).strip().upper() + for param in catalog_item.get("hyperparams", []) + if isinstance(param, dict) + } + if operator_id != "operator-builtin-shell-command" and command_keys.intersection({"COMMAND", "SCRIPT"}): + operator_image_contract_errors.append(f"{slug}: non-shell builtin must not expose COMMAND/SCRIPT") acceptance = repository_manifest.get("acceptance") if isinstance(repository_manifest.get("acceptance"), dict) else {} if not acceptance.get("requiresRealInputs"): repository_manifest_errors.append(f"{slug}: repository.json must require real inputs") @@ -204,6 +250,7 @@ def static_operator_layout() -> dict[str, Any]: "operatorFileCount": len(metadata_paths), "repositoryManifestCount": len(repository_manifests), "repositoryManifestErrors": repository_manifest_errors, + "operatorImageContractErrors": operator_image_contract_errors, "duplicateOperatorIds": sorted(duplicate_operator_ids), "duplicateOperatorNames": sorted(duplicate_operator_names), "legacyOperators": legacy, diff --git a/scripts/reconcile_system_data.py b/scripts/reconcile_system_data.py index d77aaa39..f7eba45c 100644 --- a/scripts/reconcile_system_data.py +++ b/scripts/reconcile_system_data.py @@ -110,6 +110,13 @@ def build_settings_payload(settings: JSON, k8s_endpoint: str, k8s_namespace: str def builtin_operator_patch(operator: JSON, operator_runner_image: str) -> JSON: + catalog_image = str(operator.get("image") or "").strip() + catalog_image_id = str(operator.get("imageId") or "").strip() + image = ( + catalog_image + if catalog_image and "droboticflow-operator-" in catalog_image and "droboticflow-operator-runner" not in catalog_image + else operator_runner_image + ) payload = { "name": operator["name"], "type": operator["type"], @@ -117,7 +124,11 @@ def builtin_operator_patch(operator: JSON, operator_runner_image: str) -> JSON: "description": operator["description"], "inputs": operator.get("inputs") or [], "outputs": operator.get("outputs") or [], - "image": operator_runner_image, + "image": image, + "imageId": catalog_image_id or image, + "containerCommand": operator.get("containerCommand") or [], + "containerArgs": operator.get("containerArgs") or [], + "runtimeContract": operator.get("runtimeContract") or {}, "skillDefinition": operator.get("skillDefinition") or "", "hyperparams": operator.get("hyperparams") or [], } @@ -139,7 +150,13 @@ def sync_builtin_operators(client: DroboticflowClient, operator_runner_image: st if not identifier: raise ReconcileError(f"builtin operator {operator.get('name')} is missing id") payload = builtin_operator_patch(operator, operator_runner_image) - if str(operator.get("image") or "") == operator_runner_image: + payload_image = str(payload.get("image") or "") + payload_image_id = str(payload.get("imageId") or "") + if ( + str(operator.get("image") or "") == payload_image + and str(operator.get("imageId") or "") == payload_image_id + and (operator.get("containerCommand") or []) == (payload.get("containerCommand") or []) + ): continue print(f"update operator {operator['name']}") request_json(client, "PUT", f"/api/operators/{quote_id(identifier)}", payload) diff --git a/src/lib/operators.js b/src/lib/operators.js index 7e440937..d095d24f 100644 --- a/src/lib/operators.js +++ b/src/lib/operators.js @@ -47,6 +47,9 @@ export function buildOperatorPayload(source = {}, overrides = {}) { version: trimString(overrides.version ?? source.version), image: trimString(overrides.image ?? source.image), imageId: trimString(overrides.imageId ?? source.imageId), + containerCommand: deepClone(overrides.containerCommand ?? source.containerCommand), + containerArgs: deepClone(overrides.containerArgs ?? source.containerArgs), + runtimeContract: deepClone(overrides.runtimeContract ?? source.runtimeContract), skillDefinition: trimString(overrides.skillDefinition ?? source.skillDefinition), hyperparams: deepClone(overrides.hyperparams ?? source.hyperparams ?? []), resources: deepClone(overrides.resources ?? source.resources), diff --git a/src/pages/WorkflowEditor.tsx b/src/pages/WorkflowEditor.tsx index 2b40bf8a..2f2dbe3e 100644 --- a/src/pages/WorkflowEditor.tsx +++ b/src/pages/WorkflowEditor.tsx @@ -1816,6 +1816,8 @@ export default function WorkflowEditor() { operatorVersion: op.version, operatorImageId: op.imageId || op.image, image: op.image, + containerCommand: op.containerCommand || [], + containerArgs: op.containerArgs || [], hyperparams: op.hyperparams || [], cpu: op.resources?.cpuLimit || "", memory: op.resources?.memoryLimit || "", diff --git a/src/types.ts b/src/types.ts index 5eb8b2da..7740f7b4 100644 --- a/src/types.ts +++ b/src/types.ts @@ -235,6 +235,7 @@ export interface Operator { imageId?: string; // Digest-pinned image reference when available containerCommand?: string[]; // Optional Kubernetes container command override containerArgs?: string[]; // Optional Kubernetes container args override + runtimeContract?: Record; // Runtime artifact, parameter, and acceptance contract skillDefinition?: string; // AI-readable description for workflow generation hyperparams?: HyperParam[]; // K8s Job environment variables resources?: { diff --git a/tests/backend/build_poc_images_kaniko_script_test.py b/tests/backend/build_poc_images_kaniko_script_test.py index 5b811c29..a99d1571 100644 --- a/tests/backend/build_poc_images_kaniko_script_test.py +++ b/tests/backend/build_poc_images_kaniko_script_test.py @@ -82,6 +82,9 @@ def run_script( "K8S_NAMESPACE", "OPERATOR_RUNNER_IMAGE", "POC_KANIKO_CLEAN_CONTEXT", + "POC_KANIKO_CONTEXT_SOURCE", + "POC_KANIKO_BUILD_OPERATORS", + "POC_KANIKO_OPERATOR_MANIFEST", "POC_KANIKO_IMAGE", "POC_KANIKO_GOPROXY", "POC_KANIKO_REGISTRY_MIRRORS", @@ -151,6 +154,51 @@ def test_poc_kaniko_build_creates_three_jobs_with_registry_mirror(tmp_path): assert "secretName: droboticflow-registry" in manifests +def test_poc_kaniko_build_can_create_builtin_operator_image_jobs(tmp_path): + result, calls_path, applied_dir = run_script( + tmp_path, + { + "DEPLOY_ENV": "poc", + "IMAGE_TAG": "ce9b6a1-poc-20260623135924", + "REGISTRY_HOST": "registry.example.com", + "REGISTRY_NAMESPACE": "dataflow-poc", + "POC_KANIKO_CLEAN_CONTEXT": "false", + "POC_KANIKO_BUILD_OPERATORS": "true", + }, + ) + + assert result.returncode == 0, result.stderr + result.stdout + assert "OPERATOR_IMAGE=registry.example.com/dataflow-poc/droboticflow-operator-real-source:ce9b6a1-poc-20260623135924" in result.stdout + assert "OPERATOR_IMAGE=registry.example.com/dataflow-poc/droboticflow-operator-yolo-video-target-clip:ce9b6a1-poc-20260623135924" in result.stdout + + calls = calls_path.read_text(encoding="utf-8") + assert calls.count("wait --for=condition=complete") == 99 + + manifests = "\n---\n".join(path.read_text(encoding="utf-8") for path in sorted(applied_dir.glob("*.yaml"))) + assert manifests.count("kind: Job") == 99 + assert "operator-images/builtin/real_source/Dockerfile" in manifests + assert "operator-images/builtin/yolo_video_target_clip/Dockerfile" in manifests + assert "registry.example.com/dataflow-poc/droboticflow-operator-real-source:ce9b6a1-poc-20260623135924" in manifests + assert "registry.example.com/dataflow-poc/droboticflow-operator-yolo-video-target-clip:ce9b6a1-poc-20260623135924" in manifests + + +def test_poc_kaniko_build_rejects_invalid_context_source(tmp_path): + result, calls_path, _ = run_script( + tmp_path, + { + "DEPLOY_ENV": "poc", + "IMAGE_TAG": "ce9b6a1-poc-20260623135924", + "REGISTRY_HOST": "registry.example.com", + "REGISTRY_NAMESPACE": "dataflow-poc", + "POC_KANIKO_CONTEXT_SOURCE": "dirty", + }, + ) + + assert result.returncode == 1 + assert "POC_KANIKO_CONTEXT_SOURCE must be worktree or head" in result.stderr + assert not calls_path.exists() + + def test_poc_kaniko_build_fails_fast_when_job_fails(tmp_path): result, calls_path, _ = run_script( tmp_path, diff --git a/tests/backend/operator_repository_layout_test.py b/tests/backend/operator_repository_layout_test.py index ee03fbec..635930d5 100644 --- a/tests/backend/operator_repository_layout_test.py +++ b/tests/backend/operator_repository_layout_test.py @@ -8,6 +8,7 @@ ROOT = pathlib.Path(__file__).resolve().parents[2] CATALOG_PATH = ROOT / "internal" / "api" / "builtin_operators_catalog.json" OPERATOR_ROOT = ROOT / "backend" / "operators" / "builtin" MANIFEST_SCRIPT = ROOT / "scripts" / "generate_operator_repository_manifests.py" +IMAGE_SCRIPT = ROOT / "scripts" / "generate_operator_images.py" def operator_slug(operator_id: str) -> str: @@ -72,6 +73,9 @@ class OperatorRepositoryLayoutTest(unittest.TestCase): self.assertNotIn("LEGACY_HANDLER", source) self.assertNotIn('helpers["builtin_', source) self.assertNotIn("helpers['builtin_", source) + self.assertNotIn("command_model_operator", source) + self.assertNotIn("run_command_model_operator", source) + self.assertFalse((OPERATOR_ROOT / "command_model_operator.py").exists()) def test_human_review_uses_independent_runtime_handler(self): metadata = json.loads((OPERATOR_ROOT / "human_review" / "operator.json").read_text(encoding="utf-8")) @@ -105,6 +109,14 @@ class OperatorRepositoryLayoutTest(unittest.TestCase): self.assertEqual(manifest["runtimeSource"], metadata["runtimeSource"]) self.assertEqual(manifest["runtimeHandler"], metadata["runtimeHandler"]) self.assertEqual(manifest["containerEntrypoint"], "backend/operator-runner/main.py") + self.assertTrue(manifest["image"]) + self.assertIn("droboticflow-operator-", manifest["image"]) + self.assertNotIn("droboticflow-operator-runner", manifest["image"]) + self.assertRegex(manifest["imageId"], r"@sha256:[a-f0-9]{64}$") + self.assertEqual(manifest["containerCommand"], ["python", "/app/backend/operator-runner/main.py"]) + self.assertEqual(manifest["containerArgs"], []) + self.assertTrue(manifest["runtimeContract"]["imagePolicy"]["independentFinalImage"]) + self.assertTrue(manifest["runtimeContract"]["imagePolicy"]["forbidsSharedOperatorRunnerFinalImage"]) self.assertTrue(manifest["runtimeDependencies"]) self.assertTrue(manifest["acceptance"]["requiresRealInputs"]) self.assertFalse(manifest["acceptance"]["mockOnlyAccepted"]) @@ -126,6 +138,57 @@ class OperatorRepositoryLayoutTest(unittest.TestCase): self.assertEqual(result.returncode, 0, result.stderr + result.stdout) + def test_every_builtin_operator_has_independent_image_contract(self): + catalog = json.loads(CATALOG_PATH.read_text(encoding="utf-8")) + builtin_items = [item for item in catalog if item.get("type") == "builtin"] + + self.assertEqual(len(builtin_items), 96) + for item in builtin_items: + with self.subTest(operator=item["id"]): + self.assertTrue(item.get("image")) + self.assertIn("droboticflow-operator-", item["image"]) + self.assertNotIn("droboticflow-operator-runner", item["image"]) + self.assertRegex(item.get("imageId", ""), r"@sha256:[a-f0-9]{64}$") + self.assertEqual(item.get("containerCommand"), ["python", "/app/backend/operator-runner/main.py"]) + self.assertEqual(item.get("containerArgs"), []) + contract = item.get("runtimeContract") + self.assertIsInstance(contract, dict) + self.assertTrue(contract["imagePolicy"]["independentFinalImage"]) + self.assertTrue(contract["imagePolicy"]["forbidsSharedOperatorRunnerFinalImage"]) + self.assertTrue(contract["artifacts"]["requiresLineage"]) + command_keys = { + str(param.get("key", "")).strip().upper() + for param in item.get("hyperparams", []) + } + if item["id"] == "operator-builtin-shell-command": + self.assertTrue(command_keys & {"COMMAND", "SCRIPT"}) + else: + self.assertFalse(command_keys & {"COMMAND", "SCRIPT"}) + + def test_builtin_operator_dockerfiles_are_generated(self): + catalog = json.loads(CATALOG_PATH.read_text(encoding="utf-8")) + builtin_items = [item for item in catalog if item.get("type") == "builtin"] + for item in builtin_items: + slug = operator_slug(item["id"]) + dockerfile = ROOT / "operator-images" / "builtin" / slug / "Dockerfile" + with self.subTest(operator=item["id"]): + self.assertTrue(dockerfile.exists(), f"{item['id']} must have an independent Dockerfile") + text = dockerfile.read_text(encoding="utf-8") + self.assertIn(f"DROBOTICFLOW_OPERATOR_ID={item['id']}", text) + self.assertIn(f"DROBOTICFLOW_OPERATOR_HANDLER={item['runtimeHandler']}", text) + self.assertIn('ENTRYPOINT ["python", "/app/backend/operator-runner/main.py"]', text) + + def test_operator_image_contracts_are_generated(self): + result = subprocess.run( + ["python3", str(IMAGE_SCRIPT), "--check"], + cwd=ROOT, + text=True, + capture_output=True, + check=False, + ) + + self.assertEqual(result.returncode, 0, result.stderr + result.stdout) + if __name__ == "__main__": unittest.main() diff --git a/tests/backend/poc_operator_workflow_acceptance_script_test.py b/tests/backend/poc_operator_workflow_acceptance_script_test.py index ec0c9738..3599f20e 100644 --- a/tests/backend/poc_operator_workflow_acceptance_script_test.py +++ b/tests/backend/poc_operator_workflow_acceptance_script_test.py @@ -55,7 +55,10 @@ if url.endswith("/api/health") or url.endswith("/worker-api/health"): print(json.dumps({{"status": "ok"}}, separators=(",", ":"))) elif url.endswith("/api/acceptance/poc-operator-workflows"): pathlib.Path({str(captured)!r}).write_text(body, encoding="utf-8") - print(json.dumps({{"operatorCount": 2, "runs": [{{"batch": "data-io-core", "runId": "run-1", "operatorIds": ["operator-builtin-real-source", "operator-builtin-upload"]}}]}})) + print(json.dumps({{"operatorCount": 2, "summary": {{"operatorCount": 2, "independentImageCoverage": 2, "independentImageRequired": 2, "commandWrapperOnlyOperatorCount": 0, "generatedFixtureCount": 0, "mockFixtureCount": 0}}, "runs": [{{"batch": "data-io-core", "runId": "run-1", "operatorIds": ["operator-builtin-real-source", "operator-builtin-upload"], "imageEvidence": [ + {{"operatorId": "operator-builtin-real-source", "image": "registry.example/droboticflow-operator-real-source:builtin", "imageId": "registry.example/droboticflow-operator-real-source@sha256:" + "a"*64}}, + {{"operatorId": "operator-builtin-upload", "image": "registry.example/droboticflow-operator-upload:builtin", "imageId": "registry.example/droboticflow-operator-upload@sha256:" + "b"*64}} + ]}}]}})) elif url.endswith("/worker-api/process-once"): print(json.dumps({{"processed": True}})) elif url.endswith("/api/runs/run-1"): @@ -133,7 +136,9 @@ url = next((arg for arg in args if arg.startswith("http://") or arg.startswith(" if url.endswith("/api/health") or url.endswith("/worker-api/health"): print(json.dumps({"status": "ok"}, separators=(",", ":"))) elif url.endswith("/api/acceptance/poc-operator-workflows"): - print(json.dumps({"operatorCount": 1, "runs": [{"batch": "human-gates", "runId": "run-1", "operatorIds": ["operator-builtin-human-review"]}]})) + print(json.dumps({"operatorCount": 1, "summary": {"operatorCount": 1, "independentImageCoverage": 1, "independentImageRequired": 1, "commandWrapperOnlyOperatorCount": 0, "generatedFixtureCount": 0, "mockFixtureCount": 0}, "runs": [{"batch": "human-gates", "runId": "run-1", "operatorIds": ["operator-builtin-human-review"], "imageEvidence": [ + {"operatorId": "operator-builtin-human-review", "image": "registry.example/droboticflow-operator-human-review:builtin", "imageId": "registry.example/droboticflow-operator-human-review@sha256:" + "c"*64} + ]}]})) elif url.endswith("/worker-api/process-once"): print("curl: (52) Empty reply from server", file=sys.stderr) raise SystemExit(52) @@ -208,7 +213,10 @@ url = next((arg for arg in args if arg.startswith("http://") or arg.startswith(" if url.endswith("/api/health") or url.endswith("/worker-api/health"): print(json.dumps({"status": "ok"}, separators=(",", ":"))) elif url.endswith("/api/acceptance/poc-operator-workflows"): - print(json.dumps({"operatorCount": 2, "runs": [{"batch": "external-integrations", "runId": "run-1", "operatorIds": ["operator-builtin-shell-command", "operator-builtin-mongodb-write"]}]})) + print(json.dumps({"operatorCount": 2, "summary": {"operatorCount": 2, "independentImageCoverage": 2, "independentImageRequired": 2, "commandWrapperOnlyOperatorCount": 0, "generatedFixtureCount": 0, "mockFixtureCount": 0}, "runs": [{"batch": "external-integrations", "runId": "run-1", "operatorIds": ["operator-builtin-shell-command", "operator-builtin-mongodb-write"], "imageEvidence": [ + {"operatorId": "operator-builtin-shell-command", "image": "registry.example/droboticflow-operator-shell-command:builtin", "imageId": "registry.example/droboticflow-operator-shell-command@sha256:" + "d"*64}, + {"operatorId": "operator-builtin-mongodb-write", "image": "registry.example/droboticflow-operator-mongodb-write:builtin", "imageId": "registry.example/droboticflow-operator-mongodb-write@sha256:" + "e"*64} + ]}]})) elif url.endswith("/worker-api/process-once"): print(json.dumps({"processed": True})) elif url.endswith("/api/runs/run-1"): diff --git a/tests/backend/reconcile_system_data_test.py b/tests/backend/reconcile_system_data_test.py index 46d9299e..f67073ab 100644 --- a/tests/backend/reconcile_system_data_test.py +++ b/tests/backend/reconcile_system_data_test.py @@ -103,8 +103,38 @@ class ReconcileSystemDataTest(unittest.TestCase): self.assertEqual(put_request[1], "/api/operators/operator-builtin-mongodb-write") self.assertEqual(put_request[2]["name"], "MongoDB 写入") self.assertEqual(put_request[2]["image"], "runner:new") + self.assertEqual(put_request[2]["imageId"], "runner:new") self.assertEqual(put_request[5], 6) + def test_sync_builtin_operators_preserves_independent_operator_images(self): + client = FakeClient( + { + ("GET", "/api/operators"): { + "items": [ + { + "id": "operator-builtin-real-source", + "name": "真实数据源", + "type": "builtin", + "category": "IO", + "description": "source", + "inputs": [], + "outputs": [], + "image": "registry.example/dataflow/droboticflow-operator-real-source:builtin", + "imageId": "registry.example/dataflow/droboticflow-operator-real-source@sha256:" + "a" * 64, + "containerCommand": ["python", "/app/backend/operator-runner/main.py"], + "containerArgs": [], + "runtimeContract": {"imagePolicy": {"independentFinalImage": True}}, + } + ] + }, + } + ) + + updated = reconcile.sync_builtin_operators(client, "runner:new") + + self.assertEqual(updated, 0) + self.assertEqual(len(client.requests), 1) + def test_reconcile_deletes_legacy_records_and_keeps_builtin_catalog_entries(self): client = FakeClient( {