rhaymison/Llama3-portuguese-luana-8b-instruct
rhaymison/Llama3-portuguese-luana-8b-instruct is an 8 billion parameter instruction-tuned causal language model based on Llama3, developed by rhaymison. It was fine-tuned using a superset of 290,000 Portuguese chat examples to address the scarcity of models proficient in the Portuguese language. This model is primarily optimized for chat-based applications and general conversational tasks in Portuguese, offering a specialized solution for this linguistic domain.
Loading preview...
rhaymison/Llama3-portuguese-luana-8b-instruct Overview
This model is an 8 billion parameter instruction-tuned variant of the Llama3 architecture, specifically developed by rhaymison to enhance Portuguese language capabilities. It was trained on a substantial dataset comprising 290,000 Portuguese chat examples, aiming to bridge the gap in high-quality, Portuguese-centric language models.
Key Capabilities
- Portuguese Chat Optimization: Primarily tuned for conversational tasks and chat interactions in Portuguese.
- Instruction Following: Designed to follow instructions effectively, guided by explicit prompting.
- Quantization Support: Usable with 4-bit and 8-bit quantization for reduced memory footprint, making it accessible on less powerful hardware like T4 or V100 GPUs, while the full model requires an A100.
Performance Highlights
Evaluated on the Open Portuguese LLM Leaderboard, the model demonstrates competitive performance across various Portuguese NLP tasks, achieving an average score of 68.15.
- Assin2 RTE: 89.24
- HateBR Binary: 85.93
- Assin2 STS: 72.87
- ENEM Challenge (No Images): 69
Use Cases
- Portuguese Chatbots: Ideal for developing conversational AI agents that interact in Portuguese.
- Instruction-based Tasks: Suitable for applications requiring the model to follow specific instructions in Portuguese.
- Language Understanding: Can be used for various Portuguese natural language understanding tasks, as indicated by its leaderboard performance.