gpt_train_12_128 / eval_results.json
gokulsrinivasagan's picture
End of training
3c1649d verified
{
"epoch": 0.025573961057832027,
"eval_accuracy": 0.07807518032045319,
"eval_loss": 10.078125,
"eval_runtime": 263.2075,
"eval_samples": 33767,
"eval_samples_per_second": 128.29,
"eval_steps_per_second": 2.675,
"perplexity": 23816.28811114398
}