c1_math_10d_16s / train_results.json
neginr's picture
End of training
245b958 verified
{
"epoch": 4.982278481012658,
"total_flos": 5.926379603753468e+18,
"train_loss": 0.3479239087884988,
"train_runtime": 54743.201,
"train_samples_per_second": 2.886,
"train_steps_per_second": 0.022
}