Oolel-Small-v0.1: On-Device AI for Wolof

Oolel-Small-v0.1 is a 1.5 billion parameter language model developed by soynade-research, specifically engineered for efficient operation on edge devices, such as an iPhone 12. This model integrates state-of-the-art AI with deep Wolof linguistic knowledge, trained on carefully curated high-quality data to support various general NLP tasks.

Key Capabilities

• Multilingual RAG: Supports Wolof queries with context provided in English, French, or Wolof.
• Bidirectional Translation: Facilitates seamless translation between English and Wolof.
• Wolof Text Generation: Capable of generating natural and coherent text in Wolof.
• Mathematical Problem Solving: Can handle mathematical tasks presented in Wolof.
• General NLP Tasks: Also performs well on other standard NLP tasks like summarization and text editing.

Usage Notes

Users must include a system prompt when interacting with the model. While primarily optimized for single-turn interactions, it can also be used for multi-turn conversations, though this is not its primary optimization. The model's development was led by Yaya SY (NLP Researcher) and Dioula DOUCOURE (Data & NLP Engineer).