train_math_qa_789_1760637948 / eval_results.json
rbelanec's picture
End of training
0c866e1 verified
{
"epoch": 20.0,
"eval_loss": 1.3067917823791504,
"eval_runtime": 89.6999,
"eval_samples_per_second": 66.533,
"eval_steps_per_second": 16.633,
"num_input_tokens_seen": 69309200
}