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

11 lines
311 B
JSON
Raw Normal View History

{
"ewc_loss": -85.5,
"ewc_loss_diag": 0.000370025634765625,
"ewc_loss_parallel": -0.08544921875,
"total_flos": 5.62815163329864e+19,
"train_loss": 0.0,
"train_runtime": 1304.047,
"train_samples": 125770,
"train_samples_per_second": 289.338,
"train_steps_per_second": 18.084
}