v-vasi/math-reasoning-assistant

Math Reasoning Assistant

An AI-powered math problem solver that uses step-by-step reasoning to solve grade school math problems.

About

This application uses a fine-tuned Qwen2.5-1.5B model trained via imitation learning to solve GSM8K-style math problems. The model learned to reason like GPT-4o-mini, achieving 67% accuracy on the GSM8K test set (11.7% relative improvement over the base model).

Features

• —Step-by-step mathematical reasoning
• —LaTeX rendering for equations
• —Random problem selection from GSM8K test set
• —Compare model answers with ground truth

Model

• —Base Model: Qwen/Qwen2.5-1.5B-Instruct
• —Fine-tuning Method: LoRA (Low-Rank Adaptation)
• —Teacher Model: GPT-4o-mini
• —Training Dataset: GSM8K (7,473 examples)
• —Model Card: v-vasi/qwen-math-gpt4o-imitation

Created By

Vasiliki Velona

License

Apache 2.0