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

8 lines
198 B
JSON
Raw Normal View History

{
"epoch": 3.0,
"total_flos": 6456127242240.0,
"train_loss": 1.6602046456561104,
"train_runtime": 2865.3381,
"train_samples_per_second": 3.4,
"train_steps_per_second": 0.85
}