The complete workflow for processing graphic math problems is as follows:
- Data preparation
Create a JSONL format file containing:image_urlsfield points to the local image path (e.g.file:///home/user/question.jpg)conversationsField stores the problem description (e.g. 'Find the area of the shaded part of the diagram')
- model-based reasoning
execute a commandpython scripts/inference.py --model MM-Eureka-8B --data problem.jsonl - Analysis of results
Example output:<think>识别出图形由1个矩形和2个半圆组成...</think><answer>38.5cm²</answer>
caveat: When working with complex shapes such as 3D geometries, it is recommended to use the 38B model and enable the--enable_reflectionParameters activate the reflection function.
This answer comes from the articleMM-EUREKA: A Multimodal Reinforcement Learning Tool for Exploring Visual ReasoningThe
May not be reproduced without permission:AI productivity tools " How to use MM-EUREKA to process math application problems containing pictures? What are the steps?
