Qwen/Qwen2.5-Math-7B-Instruct

Warm
Public
7.6B
FP8
131072
Sep 19, 2024
License: apache-2.0
Hugging Face
Overview

Overview

Qwen2.5-Math-7B-Instruct is part of the Qwen2.5-Math series, an upgraded collection of mathematical large language models developed by Qwen. This 7.6 billion parameter instruction-tuned model is specifically designed for solving math problems in both English and Chinese.

Key Capabilities

  • Mathematical Reasoning: Excels at solving complex mathematical problems.
  • Multilingual Support: Capable of handling math problems in both English and Chinese.
  • Reasoning Methods: Supports two primary reasoning approaches:
    • Chain-of-Thought (CoT): For step-by-step natural language reasoning.
    • Tool-integrated Reasoning (TIR): Integrates external tools for precise computation, symbolic manipulation, and algorithmic tasks, significantly improving accuracy for complex problems.
  • Performance: Achieves 85.3 on the MATH benchmark using TIR, demonstrating substantial improvements over its predecessor, Qwen2-Math.

When to Use

This model is highly recommended for applications requiring robust mathematical problem-solving capabilities, particularly those involving detailed reasoning or precise calculations. It is explicitly noted that this model series is primarily for mathematical tasks and not recommended for general-purpose applications.