b2_math_embedding / train_results.json
neginr's picture
End of training
b7d8f5e verified
{
"epoch": 5.0,
"total_flos": 3.5717155096593695e+18,
"train_loss": 0.329093080264354,
"train_runtime": 28327.3716,
"train_samples_per_second": 5.578,
"train_steps_per_second": 0.044
}