train_cb_101112_1757596151 / train_results.json
rbelanec's picture
End of training
0dce32d verified
{
"epoch": 20.0,
"num_input_tokens_seen": 621040,
"total_flos": 2.796515050979328e+16,
"train_loss": 0.20160281902148022,
"train_runtime": 249.6022,
"train_samples_per_second": 18.029,
"train_steps_per_second": 9.054
}