LLama-8B-Instruct-v0.1-MI-5e-7 / eval_results.json
tengxiao1
XT
11415f3
raw
history blame contribute delete
591 Bytes
{
"epoch": 0.9982631930527722,
"eval_logits/chosen": 0.05821159854531288,
"eval_logits/rejected": 0.049082811921834946,
"eval_logps/chosen": -0.3286878168582916,
"eval_logps/rejected": -0.3876379728317261,
"eval_loss": 1.2443650960922241,
"eval_rewards/accuracies": 0.5711382031440735,
"eval_rewards/chosen": -0.3286878168582916,
"eval_rewards/margins": 0.05895015224814415,
"eval_rewards/rejected": -0.3876379728317261,
"eval_runtime": 421.9197,
"eval_samples": 1961,
"eval_samples_per_second": 4.648,
"eval_steps_per_second": 0.292
}