初始化项目,由ModelHub XC社区提供模型
Model: Bykot/rugpt3small_perplexity_filter Source: Original Platform
This commit is contained in:
2
results/cleaning_stats.csv
Normal file
2
results/cleaning_stats.csv
Normal file
@@ -0,0 +1,2 @@
|
||||
experiment,method,source_docs,cleaned_docs,removed_docs,removed_percent,created_at_utc
|
||||
perplexity_filter,perplexity,198000,158397,39603,20.00151515151515,2026-05-18T10:25:06.185451+00:00
|
||||
|
11
results/cleaning_summary.json
Normal file
11
results/cleaning_summary.json
Normal file
@@ -0,0 +1,11 @@
|
||||
{
|
||||
"experiment": "perplexity_filter",
|
||||
"method": "perplexity",
|
||||
"base_dataset_repo": "Bykot/c4_ru_200k_split",
|
||||
"source_train_docs": 198000,
|
||||
"cleaned_docs": 158397,
|
||||
"removed_docs": 39603,
|
||||
"removed_percent": 20.00151515151515,
|
||||
"eval_docs": 2000,
|
||||
"created_at_utc": "2026-05-18T10:25:08.264865+00:00"
|
||||
}
|
||||
2
results/training_metrics.csv
Normal file
2
results/training_metrics.csv
Normal file
@@ -0,0 +1,2 @@
|
||||
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
|
||||
perplexity_filter,perplexity,ai-forever/rugpt3small_based_on_gpt2,Bykot/c4_ru_200k_split,158397,39603,2000,3.8492491245269775,46.95779053735424,3.0700321197509766,21.542594606664174,15681.1995,10.101,2026-05-18T14:53:56.568705+00:00
|
||||
|
16
results/training_metrics.json
Normal file
16
results/training_metrics.json
Normal file
@@ -0,0 +1,16 @@
|
||||
{
|
||||
"experiment": "perplexity_filter",
|
||||
"method": "perplexity",
|
||||
"model_name": "ai-forever/rugpt3small_based_on_gpt2",
|
||||
"base_dataset_repo": "Bykot/c4_ru_200k_split",
|
||||
"train_docs": 158397,
|
||||
"removed_docs": 39603,
|
||||
"eval_docs": 2000,
|
||||
"eval_loss_before": 3.8492491245269775,
|
||||
"perplexity_before": 46.95779053735424,
|
||||
"eval_loss_after": 3.0700321197509766,
|
||||
"perplexity_after": 21.542594606664174,
|
||||
"train_runtime": 15681.1995,
|
||||
"train_samples_per_second": 10.101,
|
||||
"created_at_utc": "2026-05-18T14:53:56.568705+00:00"
|
||||
}
|
||||
Reference in New Issue
Block a user