|
|
a5e747c807
|
LoRA training: artifact 1.1GB -> ~20MB adapter
|
2026-10-19 10:56:24 +08:00 |
|
|
|
b3d84ee05a
|
live-checkpoint per epoch (防僵死丢权重)
|
2026-10-16 00:24:45 +08:00 |
|
|
|
128ff308a5
|
explicit model classes + exc tail to termination-log
|
2026-10-15 16:35:55 +08:00 |
|
|
|
47427245b2
|
jev v0.2 dataset bundle + training script
|
2026-10-15 16:07:14 +08:00 |
|