UoM-CS-NeuroSymbolicAI/qwen3vl_think_math_10k
UoM-CS-NeuroSymbolicAI's qwen3vl_think_math_10k is an 8 billion parameter multimodal large language model, fine-tuned from Qwen3-VL-8B-Thinking. This model is specifically optimized for mathematical reasoning tasks, leveraging the math_interleave dataset. It aims to enhance the model's ability to process and solve complex mathematical problems, making it suitable for applications requiring advanced numerical and logical understanding.
Loading preview...
Model Overview
This model, qwen3vl_think_math_10k, is an 8 billion parameter multimodal large language model developed by UoM-CS-NeuroSymbolicAI. It is a fine-tuned variant of the base Qwen3-VL-8B-Thinking model, specifically adapted for enhanced mathematical reasoning capabilities.
Key Capabilities
- Mathematical Reasoning: The model has been fine-tuned on the
math_interleavedataset, indicating a specialized focus on improving performance in mathematical problem-solving and understanding. - Multimodal Foundation: Inherits the multimodal capabilities of its base Qwen3-VL-8B-Thinking model, suggesting potential for processing both text and visual information in mathematical contexts.
Training Details
The model underwent training with a learning rate of 5e-05 over 2 epochs, utilizing a total batch size of 16 across 2 GPUs. The optimizer used was AdamW with cosine learning rate scheduling and a 0.1 warmup ratio.
Intended Use Cases
This model is particularly well-suited for applications requiring robust mathematical understanding and problem-solving, such as:
- Automated math tutoring systems.
- Scientific research requiring computational assistance.
- Educational tools for complex mathematical concepts.