oh-dcft-v3.1-llama-3.1-8b / all_results.json
sedrickkeh's picture
End of training
1e46add verified
raw
history blame contribute delete
361 Bytes
{
"epoch": 2.996777029006739,
"eval_loss": 0.3670854866504669,
"eval_runtime": 284.8697,
"eval_samples_per_second": 40.355,
"eval_steps_per_second": 0.632,
"total_flos": 2140488220016640.0,
"train_loss": 0.33715206393986613,
"train_runtime": 42576.9653,
"train_samples_per_second": 15.39,
"train_steps_per_second": 0.03
}