Upload dataset_metadata.json with huggingface_hub
Browse files- dataset_metadata.json +52 -0
dataset_metadata.json
ADDED
|
@@ -0,0 +1,52 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"dataset_name": "RightNow Arabic LLM Corpus",
|
| 3 |
+
"language": "ar",
|
| 4 |
+
"source": "Multiple High-Quality Arabic Sources",
|
| 5 |
+
"processing_date": "2025-01-23T01:00:00",
|
| 6 |
+
"total_articles": 743288,
|
| 7 |
+
"format": "LLM Training Ready",
|
| 8 |
+
"description": "One of the largest high-quality Arabic text datasets for Large Language Model training and fine-tuning",
|
| 9 |
+
"created_by": "RightNow AI team",
|
| 10 |
+
"website": "https://rightnowai.co/",
|
| 11 |
+
|
| 12 |
+
"files": {
|
| 13 |
+
"jsonl": "arabic_wikipedia_cleaned.jsonl",
|
| 14 |
+
"txt": "arabic_wikipedia_cleaned.txt"
|
| 15 |
+
},
|
| 16 |
+
"quality_metrics": {
|
| 17 |
+
"total_words": 244153780,
|
| 18 |
+
"total_sentences": 12392064,
|
| 19 |
+
"unique_words": 1529064,
|
| 20 |
+
"vocabulary_richness": 0.0063,
|
| 21 |
+
"average_quality_score": 58.3,
|
| 22 |
+
"high_quality_articles": 185351
|
| 23 |
+
},
|
| 24 |
+
"statistics": {
|
| 25 |
+
"files_processed": 11616,
|
| 26 |
+
"files_failed": 0,
|
| 27 |
+
"articles_total": 1161600,
|
| 28 |
+
"articles_cleaned": 723251,
|
| 29 |
+
"articles_skipped": 438349,
|
| 30 |
+
"processing_time_hours": 0.97,
|
| 31 |
+
"data_size_original_gb": 8.7,
|
| 32 |
+
"data_size_cleaned_gb": 8.7,
|
| 33 |
+
"compression_ratio": 1.0
|
| 34 |
+
},
|
| 35 |
+
"technical_specs": {
|
| 36 |
+
"encoding": "UTF-8",
|
| 37 |
+
"format": "JSONL (JSON Lines)",
|
| 38 |
+
"min_article_length": 50,
|
| 39 |
+
"min_arabic_content_percentage": 70,
|
| 40 |
+
"processing_pipeline": [
|
| 41 |
+
"Source Collection",
|
| 42 |
+
"Content Cleaning",
|
| 43 |
+
"Quality Filtering",
|
| 44 |
+
"Length Filtering",
|
| 45 |
+
"Deduplication",
|
| 46 |
+
"Format Validation",
|
| 47 |
+
"Statistical Analysis"
|
| 48 |
+
]
|
| 49 |
+
},
|
| 50 |
+
"license": "MIT",
|
| 51 |
+
"citation": "RightNow Arabic LLM Corpus (2025). Collected by RightNow AI team"
|
| 52 |
+
}
|