16 lines
530 B
JSON
16 lines
530 B
JSON
|
{
|
||
|
|
"experiment": "minhash_dedup_08",
|
||
|
|
"method": "minhash_lsh",
|
||
|
|
"model_name": "ai-forever/rugpt3small_based_on_gpt2",
|
||
|
|
"base_dataset_repo": "Bykot/c4_ru_200k_split",
|
||
|
|
"train_docs": 197775,
|
||
|
|
"removed_docs": 225,
|
||
|
|
"eval_docs": 2000,
|
||
|
|
"eval_loss_before": 3.8492491245269775,
|
||
|
|
"perplexity_before": 46.95779053735424,
|
||
|
|
"eval_loss_after": 3.0526859760284424,
|
||
|
|
"perplexity_after": 21.172135967712943,
|
||
|
|
"train_runtime": 19308.2323,
|
||
|
|
"train_samples_per_second": 10.243,
|
||
|
|
"created_at_utc": "2026-05-12T06:41:45.751863+00:00"
|
||
|
|
}
|