d1_math_longest / all_results.json
sedrickkeh's picture
End of training
ef6001b verified
{
"epoch": 5.0,
"total_flos": 5.479593102698086e+18,
"train_loss": 0.3508631761257465,
"train_runtime": 58784.5389,
"train_samples_per_second": 2.688,
"train_steps_per_second": 0.021
}