BanglaLLM/BanglaLLama-3-8b-BnWiki-Instruct
BanglaLLM/BanglaLLama-3-8b-BnWiki-Instruct is an 8 billion parameter instruction-following causal language model developed by Abdullah Khan Zehady. Built upon Meta's LLaMA-3-8B, it features an extensive 16,000-token Bangla vocabulary and is fine-tuned on the BanglaLLM/bangla-alpaca-orca dataset. This model is specifically designed for instruction-following tasks in Bangla, offering enhanced linguistic capabilities for the language.
Loading preview...
Model Overview
BanglaLLM/BanglaLLama-3-8b-BnWiki-Instruct is an 8 billion parameter instruction-following model, developed by Abdullah Khan Zehady, specifically tailored for the Bangla language. It is built on the foundation of Meta's LLaMA-3-8B and incorporates an expanded 16,000-token Bangla vocabulary. The model has been fine-tuned using the BanglaLLM/bangla-alpaca-orca dataset, making it suitable for various NLP tasks requiring instruction adherence in Bangla.
Key Capabilities
- Instruction Following: Designed to understand and execute instructions in Bangla, making it suitable for interactive applications.
- Bangla Language Focus: Enhanced with a specialized 16,000-token Bangla vocabulary for improved performance in the language.
- Bilingual Support: Capable of processing both Bangla and English content.
- Foundation for Further Fine-tuning: Serves as a strong base model for domain-specific adaptations.
Usage Considerations
- Causal Language Modeling: While instruction-tuned, its core is Causal LM, meaning it generates text based on preceding tokens.
- Content Generation: Users should exercise caution as the model has not undergone detoxification and may generate harmful or offensive content.
Licensing
The model is released under the GNU General Public License v3.0.