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

9 lines
258 B
JSON
Raw Permalink Normal View History

{
"ewc_loss": 4.124641418457031e-05,
"total_flos": 5.62815163329864e+19,
"train_loss": 0.44794003818420614,
"train_runtime": 46891.4367,
"train_samples": 125770,
"train_samples_per_second": 8.046,
"train_steps_per_second": 0.503
}