jev-training-data/logs/train-progress.log
2026-10-15 11:35:13 +00:00

24 lines
1.1 KiB
Plaintext

[jev] device=cuda base=xlm-roberta-base
Some weights of XLMRobertaForSequenceClassification were not initialized from the model checkpoint at xlm-roberta-base and are newly initialized: ['classifier.dense.bias', 'classifier.dense.weight', 'classifier.out_proj.bias', 'classifier.out_proj.weight']
You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.
[jev] train=35848 calib=5291
[jev] ep0 step100/3360 loss=0.2693
[jev] ep0 step200/3360 loss=0.1835
[jev] ep0 step300/3360 loss=0.3982
[jev] ep0 step400/3360 loss=0.6066
[jev] ep0 step500/3360 loss=0.1670
[jev] ep0 step600/3360 loss=0.2391
[jev] ep0 step700/3360 loss=0.2325
[jev] ep0 step800/3360 loss=0.2099
[jev] ep0 step900/3360 loss=0.1652
[jev] ep0 step1000/3360 loss=0.1945
[jev] ep0 step1100/3360 loss=0.3984
[jev] epoch 0 calibration: {'auc': 0.9931, 'ece': 0.0136, 'n': 5291}
[jev] ep1 step1200/3360 loss=0.0956
[jev] ep1 step1300/3360 loss=0.0914
[jev] ep1 step1400/3360 loss=0.1300
[jev] ep1 step1500/3360 loss=0.2934
[jev] ep1 step1600/3360 loss=0.1928
[jev] ep1 step1700/3360 loss=0.1637
[jev] ep1 step1800/3360 loss=0.3047