Datasets:

Raziel1234
/

WebText-4

license: apache-2.0
task_categories:
  - text-generation
language:
  - en
  - he
  - ar
tags:
  - chemistry
  - biology
  - finance
  - legal
  - art
  - climate
  - code
  - medical
  - agent
size_categories:
  - 1B<n<10B

📦 Huge Multilingual Text Dataset

Overview

This repository contains a large-scale, multilingual dataset designed for training and evaluating advanced language models and AI agents. The dataset spans multiple domains, supports multiple languages, and is suitable for high-capacity models.

🔍 Key Characteristics

Dataset Size: 1B < n < 10B tokens
License: Apache License 2.0 (Apache-2.0)
Task Category:
- Text Generation
Languages Supported:
- English (en)
- Hebrew (he)
- Arabic (ar)

🧠 Domains & Tags

The dataset includes high-quality content across a wide range of fields:

🧪 Chemistry
🧬 Biology
💰 Finance
⚖️ Legal
🎨 Art
🌍 Climate
💻 Code / Software
🏥 Medical
🤖 Agent-oriented data

This diversity makes the dataset ideal for:

General-purpose LLM training
Multilingual reasoning
Pre-Training
Domain-specific fine-tuning
Autonomous and tool-using agents

🎯 Intended Use

Training large language models (LLMs)
Pre-training
Fine-tuning domain-aware or multilingual models
Research in NLP, AI agents, and reasoning systems
Benchmarking text-generation capabilities at scale

📜 License

This dataset is released under the Apache License 2.0, allowing commercial use, modification, and distribution with proper attribution.

⚠️ Disclaimer

The dataset may contain domain-specific information (e.g., medical, legal, financial) and is intended for research and development purposes only. It should not be used as a sole source for professional advice.

🙌 Acknowledgements

Created and curated as a large-scale effort to support open, multilingual, and multi-domain AI research.