llama-3_1-8b-tok10k-overfit-ua / train_results.json
antonpolishko's picture
Model save
83b9941 verified
{
"epoch": 3.0,
"total_flos": 5.294195477904884e+18,
"train_loss": 4.635029224514828,
"train_runtime": 4299.9756,
"train_samples": 95663,
"train_samples_per_second": 13.31,
"train_steps_per_second": 0.209
}