Model Overview

The najmahal/qwen-math-cebuano-1.5b-merged is a 1.5 billion parameter language model developed by najmahal, fine-tuned from unsloth/qwen2.5-math-1.5b-instruct-bnb-4bit. This model leverages the Qwen2 architecture and is specifically optimized for mathematical reasoning tasks.

Key Characteristics

• Architecture: Based on the Qwen2.5 model family.
• Parameter Count: 1.5 billion parameters, offering a balance between performance and computational efficiency.
• Fine-tuning: The model was fine-tuned using Unsloth and Huggingface's TRL library, which facilitated a 2x faster training process.
• Specialization: Primarily focused on mathematical instruction following and problem-solving.

Intended Use Cases

• Mathematical Applications: Ideal for tasks requiring numerical understanding, calculations, and mathematical reasoning.
• Educational Tools: Can be integrated into systems for teaching or assisting with math problems.
• Research: Suitable for exploring the capabilities of smaller, specialized models in mathematical domains.