zafstojano/Qwen2.5-3B-Instruct-RG-Math

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:3.1BQuant:BF16Ctx Length:32kPublished:May 14, 2025Architecture:Transformer Warm

The zafstojano/Qwen2.5-3B-Instruct-RG-Math model is a 3.1 billion parameter instruction-tuned language model based on the Qwen2.5 architecture, developed by zafstojano. It was specifically trained for reasoning and mathematical tasks, as detailed in the associated Reasoning Gym paper. With a context length of 32768 tokens, this model is optimized for complex problem-solving in mathematical and logical domains.

Loading preview...

Overview

This model, zafstojano/Qwen2.5-3B-Instruct-RG-Math, is a 3.1 billion parameter instruction-tuned variant of the Qwen2.5 architecture. It was developed by zafstojano and is specifically designed and trained for enhanced performance in reasoning and mathematical tasks.

Key Capabilities

  • Specialized Reasoning: Optimized for complex logical and mathematical problem-solving.
  • Instruction-Tuned: Responds effectively to instructions, making it suitable for interactive applications.
  • Extended Context: Supports a context length of 32768 tokens, allowing for processing longer and more intricate problems.

Good for

  • Mathematical Problem Solving: Ideal for applications requiring accurate calculations and step-by-step mathematical reasoning.
  • Logical Deduction: Suitable for tasks that involve complex logical inference and problem analysis.
  • Research in Reasoning: A valuable tool for researchers exploring and benchmarking reasoning capabilities in LLMs, as it was developed in conjunction with the Reasoning Gym paper and repository.