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

11 lines
328 B
JSON
Raw Normal View History

{
"ewc_loss": 0.03533089905977249,
"ewc_loss_diag": 0.0,
"ewc_loss_parallel": 3.5330900573171675e-05,
"total_flos": 5.62815163329864e+19,
"train_loss": 0.42502481960762,
"train_runtime": 45533.7805,
"train_samples": 125770,
"train_samples_per_second": 8.286,
"train_steps_per_second": 0.518
}