d1_math_shortest / train_results.json
sedrickkeh's picture
End of training
f7af426 verified
{
"epoch": 5.0,
"total_flos": 4.620475786573382e+18,
"train_loss": 0.3420306665453351,
"train_runtime": 57470.9786,
"train_samples_per_second": 2.749,
"train_steps_per_second": 0.021
}