dedup_ablation_sim_threshold_40 / train_results.json
sedrickkeh's picture
End of training
2b64a33 verified
{
"epoch": 2.9952,
"total_flos": 432583207665664.0,
"train_loss": 0.6040483728433267,
"train_runtime": 34034.1274,
"train_samples_per_second": 0.881,
"train_steps_per_second": 0.009
}