train_math_qa_101112_1760638065 / train_results.json
rbelanec's picture
End of training
4a6696c verified
{
"epoch": 20.0,
"num_input_tokens_seen": 77914328,
"total_flos": 3.508571438521123e+18,
"train_loss": 1.0935145752146587,
"train_runtime": 26420.3063,
"train_samples_per_second": 20.328,
"train_steps_per_second": 5.082
}