LLama-8B-Instruct-v0.1-MI-2e-5 / all_results.json
tengxiao1
TX
ae60326
{
"epoch": 0.9982631930527722,
"eval_logits/chosen": -0.3633301258087158,
"eval_logits/rejected": -0.37627702951431274,
"eval_logps/chosen": -0.5874707102775574,
"eval_logps/rejected": -0.7501260638237,
"eval_loss": 1.2528961896896362,
"eval_rewards/accuracies": 0.672764241695404,
"eval_rewards/chosen": -0.5874707102775574,
"eval_rewards/margins": 0.16265536844730377,
"eval_rewards/rejected": -0.7501260638237,
"eval_runtime": 421.6929,
"eval_samples": 1961,
"eval_samples_per_second": 4.65,
"eval_steps_per_second": 0.292,
"total_flos": 0.0,
"train_loss": 1.2796013329728513,
"train_runtime": 21025.6961,
"train_samples": 59876,
"train_samples_per_second": 2.848,
"train_steps_per_second": 0.022
}