100M_low_1000_8397 / eval_results.json
craa's picture
End of training
316c995 verified
raw
history blame contribute delete
269 Bytes
{
"epoch": 10.0,
"eval_accuracy": 0.3937655624889378,
"eval_loss": 3.3076210021972656,
"eval_runtime": 185.9043,
"eval_samples": 18011,
"eval_samples_per_second": 96.883,
"eval_steps_per_second": 6.057,
"perplexity": 27.320053752007436
}