dedup_ablation_sim_threshold_0 / train_results.json
sedrickkeh's picture
End of training
9961e6e verified
{
"epoch": 2.9952,
"total_flos": 456017134190592.0,
"train_loss": 0.5955827129192841,
"train_runtime": 34995.9246,
"train_samples_per_second": 0.857,
"train_steps_per_second": 0.009
}