gpt2-dpo-dutch / all_results.json
chasedreaminf's picture
End of training
54afc25 verified
raw
history blame contribute delete
768 Bytes
{
"epoch": 1.0,
"eval_logits/chosen": -3.4616706371307373,
"eval_logits/rejected": -3.767509937286377,
"eval_logps/chosen": -1947.5537109375,
"eval_logps/rejected": -1243.365478515625,
"eval_loss": 0.69362872838974,
"eval_rewards/accuracies": 0.5171641707420349,
"eval_rewards/chosen": 0.002152332803234458,
"eval_rewards/margins": 0.003774710465222597,
"eval_rewards/rejected": -0.0016223782440647483,
"eval_runtime": 207.5382,
"eval_samples": 5359,
"eval_samples_per_second": 25.822,
"eval_steps_per_second": 3.228,
"total_flos": 0.0,
"train_loss": 0.6930974917333634,
"train_runtime": 41.8675,
"train_samples": 482,
"train_samples_per_second": 11.513,
"train_steps_per_second": 1.457
}