train_math_qa_42_1760637604 / train_results.json
rbelanec's picture
End of training
10d043e verified
{
"epoch": 20.0,
"num_input_tokens_seen": 69300976,
"total_flos": 3.120591627456479e+18,
"train_loss": 0.5332632461089186,
"train_runtime": 17074.6562,
"train_samples_per_second": 27.958,
"train_steps_per_second": 6.99
}