PyTorch
llama
alignment-handbook
Generated from Trainer
MambaInLlama_0_75 / all_results.json
Junxiong Wang
add models
93df259
raw
history blame
771 Bytes
{
"epoch": 1.0,
"eval_logits/chosen": -0.5580978393554688,
"eval_logits/rejected": -0.579554557800293,
"eval_logps/chosen": -453.0203552246094,
"eval_logps/rejected": -622.4476928710938,
"eval_loss": 0.42684775590896606,
"eval_rewards/accuracies": 0.8410714268684387,
"eval_rewards/chosen": -2.0037572383880615,
"eval_rewards/margins": 1.612953782081604,
"eval_rewards/rejected": -3.616710901260376,
"eval_runtime": 203.0312,
"eval_samples": 4461,
"eval_samples_per_second": 21.972,
"eval_steps_per_second": 0.345,
"total_flos": 0.0,
"train_loss": 0.49609584714538074,
"train_runtime": 16148.8615,
"train_samples": 133368,
"train_samples_per_second": 8.259,
"train_steps_per_second": 0.258
}