Alelcv27/Qwen3-4B-INST-Math-v2
Alelcv27/Qwen3-4B-INST-Math-v2 is a 4 billion parameter Qwen3 instruction-tuned language model developed by Alelcv27, fine-tuned from unsloth/qwen3-4b-instruct-2507-unsloth-bnb-4bit. This model was specifically optimized for mathematical tasks, leveraging Unsloth and Huggingface's TRL library for faster training. It offers a 32768 token context length, making it suitable for processing longer mathematical problems and related instructions.
Loading preview...
Alelcv27/Qwen3-4B-INST-Math-v2 Overview
Alelcv27/Qwen3-4B-INST-Math-v2 is a 4 billion parameter instruction-tuned model based on the Qwen3 architecture, developed by Alelcv27. It was fine-tuned from unsloth/qwen3-4b-instruct-2507-unsloth-bnb-4bit with a specific focus on enhancing its capabilities in mathematical reasoning and problem-solving. The training process utilized Unsloth and Huggingface's TRL library, which enabled a 2x faster fine-tuning compared to standard methods.
Key Capabilities
- Mathematical Proficiency: Specifically optimized for handling mathematical instructions and tasks.
- Efficient Training: Benefits from Unsloth's accelerated training techniques, making it a product of efficient fine-tuning.
- Qwen3 Architecture: Leverages the robust base of the Qwen3 model family.
- Extended Context: Supports a 32768 token context length, allowing for more complex and multi-step mathematical problems.
Good For
- Applications requiring strong mathematical understanding and problem-solving.
- Tasks involving numerical reasoning, equations, and logical mathematical sequences.
- Developers looking for a compact yet capable model for math-centric AI applications.