๐ฌ Chat as a way to query SQL! The Airtrain AI team is happy to share a new Hugging Face Space that lets you interact with Hugging Face Hub datasets using a natural language chatbot. ๐ค
This Space is forked from davidberenstein1957/text-to-sql-hub-datasetsย byย @davidberenstein1957 and features chat capability with improved table naming. The tool works with Hugging Faceโs recently released in-browser DuckDB-based SQL query engine for datasets.
reacted to Salama1429's
post with ๐5 months ago
๐ Introducing the 101 Billion Arabic Words Dataset
๐ Exciting Milestone in Arabic Language Technology! hashtag#NLP hashtag#ArabicLLM hashtag#LanguageModels
๐ Why It Matters: 1. ๐ Large Language Models (LLMs) have brought transformative changes, primarily in English. It's time for Arabic to shine! 2. ๐ฏ This project addresses the critical challenge of bias in Arabic LLMs due to reliance on translated datasets.
๐ Approach: 1. ๐ช Undertook a massive data mining initiative focusing exclusively on Arabic from Common Crawl WET files. 2. ๐งน Employed state-of-the-art cleaning and deduplication processes to maintain data quality and uniqueness.
๐ Impact: 1. ๐ Created the largest Arabic dataset to date with 101 billion words. 2. ๐ Enables the development of Arabic LLMs that are linguistically and culturally accurate. 3. ๐ Sets a global benchmark for future Arabic language research.