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

11 lines
324 B
JSON
Raw Permalink Normal View History

{
"ewc_loss": 0.041259765625,
"ewc_loss_diag": 4.124641418457031e-05,
"ewc_loss_parallel": 0.0,
"total_flos": 5.62815163329864e+19,
"train_loss": 0.4491925383311204,
"train_runtime": 45746.0141,
"train_samples": 125770,
"train_samples_per_second": 8.248,
"train_steps_per_second": 0.516
}