Files
Llama-3.1-8B-Instruct_SFT_s…/all_results.json

9 lines
259 B
JSON
Raw Permalink Normal View History

{
"ewc_loss": 0.0002346038818359375,
"total_flos": 3.1829322264618205e+19,
"train_loss": 0.9308701857364808,
"train_runtime": 37091.194,
"train_samples": 145693,
"train_samples_per_second": 11.784,
"train_steps_per_second": 0.737
}