Upload README.md with huggingface_hub

README.md

@@ -7,21 +7,21 @@ language:
 size_categories:
   - 1M<n<10M
 tags:
-  - chain-of-thought
-  - reasoning
   - diverse
   - curated
   - deduplication
+  - multi-domain
   - stem
   - legal
   - scientific
   - encyclopedic
+  - source-text
 configs:
   - config_name: default
     data_files:
       - split: train
         path: cot_diverse_2.5m.parquet
-pretty_name: Diverse CoT Source Dataset
+pretty_name: Diverse Source Text Dataset (2.5M)
 dataset_info:
   features:
     - name: text
@@ -39,33 +39,34 @@ dataset_info:
       num_examples: 2500000
 ---
 
-# Diverse CoT Source Dataset (2.5M)
+# Diverse Source Text Dataset (2.5M)
 
-A curated, deduplicated, multi-domain English text dataset designed as source material for synthetic chain-of-thought (CoT) generation. The dataset blends 7 sources across STEM, legal, scientific, encyclopedic, Q&A, and general knowledge domains to maximize reasoning diversity.
+A curated, deduplicated, multi-domain English text dataset blending 7 sources across STEM, legal, scientific, encyclopedic, Q&A, and general knowledge domains. Designed as high-quality, diverse source material for downstream NLP tasks such as synthetic data generation, fine-tuning, and text analysis.
 
 ## Dataset Summary
 
 | | |
 |---|---|
 | **Total samples** | 2,500,000 |
+| **Estimated tokens** | ~2.8B (GPT-2) / ~2.4B (modern tokenizers) |
 | **Language** | English |
 | **Format** | Parquet (ZSTD compressed) |
 | **File size** | 4.28 GB |
 | **Text length** | 200 - 50,000 characters |
-| **Mean length** | 4,656 characters |
+| **Mean length** | 4,656 characters (~1,107 tokens) |
 | **Median length** | 2,439 characters |
 
 ## Source Breakdown
 
-| Source | Samples | Share | Avg Chars | Quality Score | Domain |
-|--------|--------:|------:|----------:|--------------:|--------|
-| FineWeb EDU (broad, 3.0-4.0) | 750,000 | 30% | 4,997 | 3.39 | General educational |
-| DCLM-baseline | 500,000 | 20% | 2,295 | 0.89 | Commonsense / explanatory |
-| FineWeb EDU (high, >= 4.0) | 375,000 | 15% | 4,923 | 4.18 | STEM / high-quality educational |
-| Pile - FreeLaw | 250,000 | 10% | 14,458 | N/A | Legal (court opinions, filings) |
-| Pile - PubMed Abstracts | 250,000 | 10% | 1,335 | N/A | Biomedical / scientific |
-| Pile - StackExchange | 200,000 | 8% | 2,190 | N/A | Technical Q&A |
-| Pile - Wikipedia (en) | 175,000 | 7% | 2,923 | N/A | Encyclopedic |
+| Source | Samples | Share | Avg Chars | Avg Tok/Doc | Quality Score | Domain |
+|--------|--------:|------:|----------:|------------:|--------------:|--------|
+| FineWeb EDU (broad, 3.0-4.0) | 750,000 | 30% | 4,997 | 1,063 | 3.39 | General educational |
+| DCLM-baseline | 500,000 | 20% | 2,295 | 572 | 0.89 | Commonsense / explanatory |
+| FineWeb EDU (high, >= 4.0) | 375,000 | 15% | 4,923 | 1,023 | 4.18 | STEM / high-quality educational |
+| Pile - FreeLaw | 250,000 | 10% | 14,458 | 3,781 | N/A | Legal (court opinions, filings) |
+| Pile - PubMed Abstracts | 250,000 | 10% | 1,335 | 292 | N/A | Biomedical / scientific |
+| Pile - StackExchange | 200,000 | 8% | 2,190 | 761 | N/A | Technical Q&A |
+| Pile - Wikipedia (en) | 175,000 | 7% | 2,923 | 685 | N/A | Encyclopedic |
 
 ## Schema
 
@@ -105,7 +106,7 @@ Total removed: 93,069 / 3,000,000 (3.1%)
 ```python
 from datasets import load_dataset
 
-ds = load_dataset("blythet/cot-diverse-2.5m", split="train")
+ds = load_dataset("blythet/diverse-2.5m", split="train")
 print(ds)
 # Dataset({
 #     features: ['text', 'id', 'url', 'source', 'quality_score'],
@@ -121,14 +122,14 @@ high_quality = ds.filter(lambda x: x["quality_score"] is not None and x["quality
 
 ## Intended Use
 
-This dataset is designed as **input material for synthetic chain-of-thought generation** using large language models. The domain diversity ensures the resulting CoT data covers:
+This dataset provides high-quality, diverse English text suitable for:
 
-- STEM reasoning and mathematical explanations
-- Legal analysis and case law interpretation
-- Scientific literature comprehension
-- Technical problem-solving (StackExchange)
-- General knowledge and encyclopedic reasoning
-- Everyday commonsense explanations
+- Synthetic data generation (e.g., chain-of-thought, instruction tuning)
+- Fine-tuning language models across multiple domains
+- Text analysis and NLP research
+- Domain-specific data extraction (legal, scientific, educational, technical)
+
+The domain diversity covers STEM, legal reasoning, scientific literature, technical Q&A, encyclopedic knowledge, and general commonsense explanations.
 
 ## Limitations
 
@@ -147,11 +148,11 @@ This dataset is released under **ODC-By** (Open Data Commons Attribution License
 ## Citation
 
 ```bibtex
-@dataset{cot_diverse_2.5m,
-  title={Diverse CoT Source Dataset},
+@dataset{diverse_2.5m,
+  title={Diverse Source Text Dataset},
   author={blythet},
   year={2025},
-  url={https://huggingface.co/datasets/blythet/cot-diverse-2.5m},
-  note={2.5M curated, deduplicated multi-domain English texts for chain-of-thought generation}
+  url={https://huggingface.co/datasets/blythet/diverse-2.5m},
+  note={2.5M curated, deduplicated multi-domain English texts}
 }
 ```