TokenAI open sources Horus-1.0 training code

TokenAI's move to open-source the training code for a region-specific model is a significant step for AI democratization in the Middle East.

• –Horus-1.0-4B is the first open-source LLM trained from scratch in Egypt, filling a major gap in localized model support.
• –The model reportedly outperforms larger counterparts like Llama 3.1-8B on specific benchmarks while maintaining a smaller footprint.
• –By releasing the training scripts under the MIT License, TokenAI enables developers to build and fine-tune localized models for specific regional needs.
• –The project includes a full ecosystem with the neuralnode framework and Replica TTS, suggesting a move towards a complete indigenous AI stack.
• –The 4B parameter size hits a "sweet spot" for edge deployment and affordable fine-tuning for individual researchers.