c1_math_10d_1s / all_results.json
neginr's picture
End of training
0e2cadf verified
{
"epoch": 4.982278481012658,
"total_flos": 5.925003403780555e+18,
"train_loss": 0.36658277566113123,
"train_runtime": 54896.0887,
"train_samples_per_second": 2.878,
"train_steps_per_second": 0.022
}