reasoning_hp_ablations_bsz96 / all_results.json
sedrickkeh's picture
End of training
fec8b12 verified
{
"epoch": 2.9991577765300392,
"total_flos": 4618853558419456.0,
"train_loss": 0.436508960367019,
"train_runtime": 120897.2745,
"train_samples_per_second": 2.828,
"train_steps_per_second": 0.029
}