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

11 lines
348 B
JSON
Raw Normal View History

{
"ewc_loss": 0.08365151286125183,
"ewc_loss_diag": 3.790855407714844e-05,
"ewc_loss_parallel": 0.00045809714356437325,
"total_flos": 5.62815163329864e+19,
"train_loss": 0.4756091437999748,
"train_runtime": 47393.4997,
"train_samples": 125770,
"train_samples_per_second": 7.961,
"train_steps_per_second": 0.498
}