hp_ablations_qwen_epoch4 / train_results.json
sedrickkeh's picture
End of training
e6c7e36 verified
raw
history blame contribute delete
220 Bytes
{
"epoch": 3.993162393162393,
"total_flos": 3673450900094976.0,
"train_loss": 0.5887784766688194,
"train_runtime": 47860.2816,
"train_samples_per_second": 18.773,
"train_steps_per_second": 0.037
}