Qwen/Qwen2.5-Math-1.5B-Instruct

Warm
Public
1.5B
BF16
131072
License: apache-2.0
Hugging Face
Overview

Qwen2.5-Math-1.5B-Instruct: Specialized Mathematical LLM

Qwen2.5-Math-1.5B-Instruct is part of the Qwen2.5-Math series, an upgrade from the earlier Qwen2-Math models, developed by Qwen. This 1.5 billion parameter instruction-tuned model is specifically engineered for solving mathematical problems in both English and Chinese.

Key Capabilities

  • Mathematical Problem Solving: Primarily supports solving math problems through Chain-of-Thought (CoT) and Tool-integrated Reasoning (TIR).
  • Multilingual Support: Expanded to handle mathematical tasks in both Chinese and English.
  • Enhanced Reasoning: Achieves significant performance improvements on Chinese and English mathematics benchmarks using CoT compared to its predecessor.
  • Computational Accuracy: TIR further improves the model's proficiency in precise computation, symbolic manipulation, and algorithmic manipulation, addressing challenges faced by CoT alone.

Good For

  • Developers requiring an LLM specifically for mathematical problem-solving.
  • Applications needing step-by-step reasoning (CoT) or tool-assisted computation (TIR) for math.
  • Use cases involving mathematical tasks in both English and Chinese.