qlora-llama3b-iterative / all_results.json
pakphum's picture
End of training
9823daf verified
raw
history blame
363 Bytes
{
"epoch": 4.444444444444445,
"eval_loss": 0.005105508491396904,
"eval_runtime": 8.0865,
"eval_samples_per_second": 12.366,
"eval_steps_per_second": 12.366,
"total_flos": 1.038706354200576e+16,
"train_loss": 0.1301481350355316,
"train_runtime": 1611.4177,
"train_samples_per_second": 2.482,
"train_steps_per_second": 0.31
}