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

11 lines
335 B
JSON
Raw Normal View History

{
"ewc_loss": -0.7421875,
"ewc_loss_diag": 2.944469451904297e-05,
"ewc_loss_parallel": -0.00775146484375,
"total_flos": 5.62815163329864e+19,
"train_loss": 0.35840998991934647,
"train_runtime": 47115.5066,
"train_samples": 125770,
"train_samples_per_second": 8.008,
"train_steps_per_second": 0.501
}