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

11 lines
317 B
JSON
Raw Permalink Normal View History

{
"ewc_loss": 0.052088044583797455,
"ewc_loss_diag": 0.0,
"ewc_loss_parallel": 2.6044022888527252e-05,
"total_flos": 5.62815163329864e+19,
"train_loss": 0.0,
"train_runtime": 852.4232,
"train_samples": 125770,
"train_samples_per_second": 442.632,
"train_steps_per_second": 27.666
}