train_math_qa_456_1760637840 / eval_results.json
rbelanec's picture
End of training
4f908e3 verified
{
"epoch": 20.0,
"eval_loss": 0.675398588180542,
"eval_runtime": 45.1802,
"eval_samples_per_second": 66.047,
"eval_steps_per_second": 16.512,
"num_input_tokens_seen": 77891968
}