b2_math_length / train_results.json
neginr's picture
End of training
9f28305 verified
{
"epoch": 4.982278481012658,
"total_flos": 6.179014728805941e+18,
"train_loss": 0.35052244286469325,
"train_runtime": 55060.6716,
"train_samples_per_second": 2.87,
"train_steps_per_second": 0.022
}