dedup_ablation_sim_threshold_none / train_results.json
sedrickkeh's picture
End of training
96e2620 verified
{
"epoch": 2.9952,
"total_flos": 7.00883763227263e+17,
"train_loss": 0.5766575982173284,
"train_runtime": 33757.7333,
"train_samples_per_second": 0.889,
"train_steps_per_second": 0.009
}