llama3-L1-SFT-L2-KTO / all_results.json
EllieS's picture
End of training
1246b45 verified
raw
history blame contribute delete
764 Bytes
{
"epoch": 0.9998751404669747,
"eval_logits/chosen": 0.19315297901630402,
"eval_logits/rejected": 1.0988490581512451,
"eval_logps/chosen": -12.085772514343262,
"eval_logps/rejected": -1018.5144653320312,
"eval_loss": 0.2121797353029251,
"eval_rewards/accuracies": 1.0,
"eval_rewards/chosen": 0.32569777965545654,
"eval_rewards/margins": 9.880459785461426,
"eval_rewards/rejected": -9.554760932922363,
"eval_runtime": 0.4362,
"eval_samples": 5,
"eval_samples_per_second": 11.464,
"eval_steps_per_second": 6.878,
"total_flos": 0.0,
"train_loss": 0.2449444185841929,
"train_runtime": 3711.3289,
"train_samples": 16017,
"train_samples_per_second": 4.316,
"train_steps_per_second": 0.539
}