llama-7b-dpo-qlora-relu / all_results.json
chanchan7's picture
Model save
7221958 verified
{
"epoch": 1.0,
"eval_logits/chosen": 0.14104285836219788,
"eval_logits/rejected": 0.23967117071151733,
"eval_logps/chosen": -3150.4404296875,
"eval_logps/rejected": -2686.59619140625,
"eval_loss": 0.6422629356384277,
"eval_rewards/accuracies": 0.6669999957084656,
"eval_rewards/chosen": 0.9449260830879211,
"eval_rewards/margins": 0.30469661951065063,
"eval_rewards/rejected": 0.6402294039726257,
"eval_runtime": 435.1825,
"eval_samples": 2000,
"eval_samples_per_second": 4.596,
"eval_steps_per_second": 1.149,
"train_loss": 0.6582325137556925,
"train_runtime": 111789.0929,
"train_samples": 61135,
"train_samples_per_second": 0.547,
"train_steps_per_second": 0.034
}