rhaymison/Mistral-portuguese-luana-7b Overview

The rhaymison/Mistral-portuguese-luana-7b is a 7 billion parameter model, fine-tuned from the Mistral 7B base model. Its primary distinction is its extensive training on a superset of 200,000 instructions exclusively in Portuguese, aiming to fill a significant gap in Portuguese-language models. This specialization makes it particularly adept at understanding and executing instructional tasks in Portuguese.

Key Capabilities & Features

• Portuguese Language Focus: Tuned specifically for Portuguese, offering improved performance for tasks in this language.
• Instruction Following: Optimized for instructional tasks, making it suitable for applications requiring precise responses to commands or queries.
• Quantization Support: Compatible with 4-bit and 8-bit quantization, allowing deployment on less powerful hardware like T4 or V100 GPUs, while the full model requires an A100.
• GGUF Compatibility: Available in GGUF formats (e.g., Mistral-portuguese-luana-7b-q8-gguf, Mistral-portuguese-luana-7b-f16-gguf) for use with LlamaCpp, enhancing deployment flexibility.
• Leaderboard Performance: Achieves an average score of 64.27 on the Open Portuguese LLM Leaderboard, with notable scores in tasks like Assin2 RTE (90.63) and HateBR Binary (77.24).

When to Use This Model

This model is ideal for developers and researchers focused on:

• Applications requiring strong performance in Portuguese natural language processing.
• Instruction-following tasks where the model needs to act or behave according to specific prompts.
• Projects where resource efficiency is important, leveraging its 4-bit and 8-bit quantization options.
• Use cases benefiting from a model specifically trained to understand and generate Portuguese text effectively.