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

8 lines
201 B
JSON
Raw Permalink Normal View History

{
"epoch": 3.0,
"total_flos": 7768633466880.0,
"train_loss": 1.6535469640083522,
"train_runtime": 3364.7928,
"train_samples_per_second": 3.453,
"train_steps_per_second": 0.864
}