no_pipeline_math_1k / train_results.json
neginr's picture
End of training
8dce93b verified
{
"epoch": 6.666666666666667,
"total_flos": 7.211721962605773e+16,
"train_loss": 0.3436877076114927,
"train_runtime": 2253.009,
"train_samples_per_second": 3.107,
"train_steps_per_second": 0.031
}