Files
Llama3.2-3B_Paper_Impact_SFT/all_results.json

12 lines
349 B
JSON
Raw Normal View History

{
"epoch": 3.0,
"eval_loss": 0.1446007937192917,
"eval_runtime": 1173.577,
"eval_samples_per_second": 54.71,
"eval_steps_per_second": 1.71,
"total_flos": 6.274047864041636e+18,
"train_loss": 0.03664169063040653,
"train_runtime": 42810.6565,
"train_samples_per_second": 6.203,
"train_steps_per_second": 0.048
}