Lugha-Llama/Lugha-Llama-8B-wura
Lugha-Llama/Lugha-Llama-8B-wura is an 8 billion parameter language model developed by Happy Buzaaba, Alexander Wettig, David Ifeoluwa Adelani, and Christiane Fellbaum, based on Llama-3.1-8B with a 32768-token context length. This model is specifically adapted for African languages, addressing their underrepresentation in large language model training datasets. It achieves leading performance among open-source models on African language benchmarks like IrokoBench and AfriQA, making it ideal for applications requiring strong understanding and generation in these languages.
Loading preview...
Lugha-Llama-8B-wura: African Language Adaptation
Lugha-Llama-8B-wura is an 8 billion parameter model derived from Llama-3.1-8B, developed by Happy Buzaaba, Alexander Wettig, David Ifeoluwa Adelani, and Christiane Fellbaum. It addresses the challenge of underrepresentation of low-resource African languages in mainstream large language models.
Key Capabilities
- African Language Proficiency: Specifically adapted to improve performance on various African languages.
- Benchmark Leading Performance: Achieves the best performance among open-source models on challenging African language benchmarks, including IrokoBench and AfriQA.
- Cross-lingual Question Answering: Excels in cross-lingual open-retrieval question answering for African languages.
- Context Length: Features a substantial context window of 32768 tokens.
Good For
- Applications targeting African languages: Ideal for developers building solutions that require robust understanding and generation in various African languages.
- Research in low-resource NLP: Provides a strong baseline and tool for researchers working on natural language processing for underrepresented languages.
- Improving accessibility: Can be used to develop tools and services that make information more accessible to speakers of African languages.