d1_math_gpt / train_results.json
neginr's picture
End of training
696fb4c verified
raw
history blame contribute delete
208 Bytes
{
"epoch": 5.0,
"total_flos": 4.66034976231732e+18,
"train_loss": 0.30737045316802347,
"train_runtime": 28730.5073,
"train_samples_per_second": 5.499,
"train_steps_per_second": 0.043
}