Files
llama3-8b-full-pretrain-was…/train_results.json

8 lines
202 B
JSON
Raw Permalink Normal View History

{
"epoch": 3.0,
"total_flos": 12614518824960.0,
"train_loss": 1.6530290696965189,
"train_runtime": 5359.1808,
"train_samples_per_second": 3.517,
"train_steps_per_second": 0.879
}