Showing cs.CVShow all
2 papers · 1 filter
cs.CV2025
CapGeo: A Caption-Assisted Approach to Geometric Reasoning
Yuying Li, Siyi Qian, Hao Liang +4
Geometric reasoning remains a core challenge for Multimodal Large Language Models (MLLMs). Even the most advanced closed-source systems, such as GPT-O3 and Gemini-2.5-Pro, still st…
cs.CV2025
Unlocking the Potential of Difficulty Prior in RL-based Multimodal Reasoning
Mingrui Chen, Haogeng Liu, Hao Liang +3
In this work, we investigate how explicitly modeling problem's difficulty prior information shapes the effectiveness of reinforcement learning based fine-tuning for multimodal reas…