434 B
434 B
| 1 | experiment | method | model_name | base_dataset_repo | train_docs | removed_docs | eval_docs | eval_loss_before | perplexity_before | eval_loss_after | perplexity_after | train_runtime | train_samples_per_second | created_at_utc |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2 | minhash_dedup_08 | minhash_lsh | ai-forever/rugpt3small_based_on_gpt2 | Bykot/c4_ru_200k_split | 197775 | 225 | 2000 | 3.8492491245269775 | 46.95779053735424 | 3.0526859760284424 | 21.172135967712943 | 19308.2323 | 10.243 | 2026-05-12T06:41:45.751863+00:00 |