Datasets:
๐ฎ๐ณ SOVEREIGN INDIAN INTELLIGENCE ๐ฎ๐ณ
๐ฅ SKT AI LABS ๐ฅ
๐ฎ๐ณ MADE IN BHARAT ๐งฌ ST-CORE-TOKENS LOGIC DISTILLED 588 GB+
๐ฑ SKT-Ai-Labs/ST-CORE-TOKENS
ST-CORE-TOKENS ek ultra-refined, high-density tokenized dataset hai jo SKT AI LABS dwara develop kiya gaya hai. Yeh dataset Indian LLMs ki Cognitive Logic aur Advanced Reasoning ko empower karne ke liye design kiya gaya hai.
๐ฐ๏ธ Dataset Details
| Attribute | Details |
|---|---|
| Curated by | SKT AI LABS |
| Architecture | Proprietary SKT-Logic-MoE Compatible |
| Data Type | Ultra High-Density Distilled Tokens |
| Languages | English + Context-Aware Hinglish |
| Total Shards | 1,400+ Parquet Files |
| Size | 588 GB+ |
๐ฑ Key Features
- Pure Logic Distillation โ Junk aur noise data ko eliminate karke sirf high-reasoning tokens extract kiye gaye hain.
- Cognitive Boost โ Model ki zero-shot reasoning aur problem-solving capabilities ko enhance karta hai.
- Sovereign Bharat Focus โ Indian cultural nuances aur technical Hinglish logic ko priority di gayi hai.
- MoE Optimized โ Project Surya jaise Mixture-of-Experts architectures ke liye optimized expert routing.
๐ Uses
- Foundation Pre-training: Naye models ko logic-rich base dene ke liye.
- Fine-Tuning: Existing models ki halluination kam karne aur reasoning sudhaarne ke liye.
- Research: Sovereign AI development ke liye standard logic corpus.
Developed with โค๏ธ by SKT AI LABS Empowering Bharat with Sovereign Intelligence ๐ฎ๐ณ
Citation
@misc{skt-ai-labs-2026-st-core-tokens,
title = {ST-CORE-TOKENS: High-Density Logic Distilled Dataset for Sovereign Indian LLMs},
author = {SKT AI LABS},
year = {2026},
publisher = {Hugging Face}
}
- Downloads last month
- 18