Update README.md

README.md

@@ -1,3 +1,54 @@
----
-license: mit
----
+---
+dataset_info:
+  features:
+  - name: text
+    dtype: string
+  splits:
+  - name: train
+    num_bytes: 20000000000
+    num_examples: 20000000
+  download_size: 20000000000
+  dataset_size: 20000000000
+license: mit
+language:
+- si
+tags:
+- sinhala
+- clean-corpus
+- low-resource-languages
+- nlp
+- embeddings
+- remeinium
+pretty_name: Cleaned Sinhala Text Corpus
+---
+
+# 📚 Cleaned Sinhala Text Corpus
+
+The **Cleaned Sinhala Text Corpus** is a large-scale, high-quality Sinhala text dataset prepared and released by **Remeinium Corp**.  
+
+This dataset contains **20GB** of Sinhala-only text (20 × 1GB `.gz` files when uncompressed), carefully cleaned and tokenized.  
+It is designed for **language modeling, embeddings, and NLP research**.
+
+---
+
+## ✨ Key Features
+- **Language**: Sinhala (si)
+- **Size**: ~20GB uncompressed (20 × 1GB files)
+- **Format**: Plain text (`.gz` compressed), one sentence per line
+- **Cleaning**:
+  - Removed all non-Sinhala characters
+  - No punctuation, numbers, or symbols
+  - Deduplicated and normalized
+- **Tokenization**: Pre-tokenized with consistent whitespace separation
+
+---
+
+## 🔧 Usage
+
+You can load this dataset with `datasets`:
+
+```python
+from datasets import load_dataset
+
+dataset = load_dataset("Remeinium/Cleaned_Sinhala_Text")
+print(dataset["train"][0])