PyTorch
llama
alignment-handbook
Generated from Trainer
Mamba2InLlama_0_875 / all_results.json
Junxiong Wang
add models
e5a6ffa
raw
history blame contribute delete
769 Bytes
{
"epoch": 1.0,
"eval_logits/chosen": 0.3841487765312195,
"eval_logits/rejected": 0.3401743471622467,
"eval_logps/chosen": -458.9393005371094,
"eval_logps/rejected": -584.5836791992188,
"eval_loss": 0.4761734902858734,
"eval_rewards/accuracies": 0.7946428656578064,
"eval_rewards/chosen": -1.4026715755462646,
"eval_rewards/margins": 1.1953023672103882,
"eval_rewards/rejected": -2.5979740619659424,
"eval_runtime": 172.3438,
"eval_samples": 4461,
"eval_samples_per_second": 25.884,
"eval_steps_per_second": 0.406,
"total_flos": 0.0,
"train_loss": 0.5273771832863338,
"train_runtime": 14157.4064,
"train_samples": 133368,
"train_samples_per_second": 9.42,
"train_steps_per_second": 0.294
}