reasoning_hp_ablations_bsz32 / all_results.json
sedrickkeh's picture
End of training
accd6ca verified
raw
history blame
206 Bytes
{
"epoch": 3.0,
"total_flos": 4620716816613376.0,
"train_loss": 0.39878227835304614,
"train_runtime": 132014.4933,
"train_samples_per_second": 2.59,
"train_steps_per_second": 0.081
}