espressovi/BODHI-qwen-2.5-32b-distil
The espressovi/BODHI-qwen-2.5-32b-distil model is a specialized language model created by espressovi, derived from Qwen/Qwen2.5-32B through Long-CoT distillation. This model is specifically optimized for mathematical reasoning tasks. It demonstrates strong performance on benchmarks like AIME25, making it suitable for applications requiring advanced mathematical problem-solving capabilities.
Loading preview...
Overview
The espressovi/BODHI-qwen-2.5-32b-distil model is an artifact from the BODHI project, developed by espressovi. It is a specialized language model created through a process called Long-CoT distillation, starting from the larger Qwen/Qwen2.5-32B base model. This distillation process has focused its capabilities, particularly enhancing its performance in mathematical domains.
Key Capabilities
- Mathematical Reasoning: The primary focus of this model is on mathematical problem-solving. It is designed to excel in tasks requiring complex calculations and logical deduction within a mathematical context.
- Distilled Performance: By using Long-CoT distillation, the model aims to retain high performance in its specialized area while potentially offering efficiencies over its larger base model.
Performance Highlights
- AIME25 Benchmark: The model achieves a performance of 59.23% on the AIME25 benchmark, evaluated at T=1.0, with 16K tokens and pass@8. This indicates a strong capability in solving advanced mathematics problems.
When to Use This Model
This model is particularly well-suited for use cases that demand high accuracy and proficiency in mathematical reasoning. If your application involves solving complex math problems, competitive programming challenges, or educational tools requiring mathematical understanding, this model offers a specialized solution.