train_piqa_1755545132 / train_results.json
rbelanec's picture
End of training
dfded16 verified
{
"epoch": 10.0,
"num_input_tokens_seen": 22103448,
"total_flos": 9.953082733888143e+17,
"train_loss": 0.5287332604767141,
"train_runtime": 6256.026,
"train_samples_per_second": 23.179,
"train_steps_per_second": 5.796
}