hp_ablations_qwen_bsz1024 / all_results.json
sedrickkeh's picture
End of training
2c40a07 verified
raw
history blame contribute delete
362 Bytes
{
"epoch": 2.994871794871795,
"eval_loss": 0.6190248131752014,
"eval_runtime": 222.7123,
"eval_samples_per_second": 53.086,
"eval_steps_per_second": 0.418,
"total_flos": 2754957111459840.0,
"train_loss": 0.6167326612196739,
"train_runtime": 35776.9525,
"train_samples_per_second": 18.835,
"train_steps_per_second": 0.018
}