reasoning_hp_ablations_bsz128 / train_results.json
sedrickkeh's picture
End of training
e9c6555 verified
raw
history blame
219 Bytes
{
"epoch": 2.998877035373386,
"total_flos": 4617526237986816.0,
"train_loss": 0.4445273021968563,
"train_runtime": 119673.128,
"train_samples_per_second": 2.857,
"train_steps_per_second": 0.022
}