Search papers, labs, and topics across Lattice.
Mags-RL is introduced, an agentic reinforcement learning framework that equips MLLMs with a super-resolution "magnifying glass" agent for fine-grained visual inspection. The MLLM first generates an initial rationale and identifies regions of interest, then uses the super-resolution agent to upscale those regions before refining its reasoning. A novel curriculum learning strategy enables data-efficient RL training, achieving strong performance with only 40 training samples.
Mags-RL lets multimodal LLMs see the forest *and* the trees, using reinforcement learning to guide a super-resolution agent that selectively enhances image regions for improved reasoning without extra annotations.
Despite their popularity and success, Multimodal Large Language Models (MLLMs) often struggle to interpret images accurately, which limits their reasoning capability in complex scenarios (e.g., high object density and complex background clutter). Prior work mainly addresses this limitation by incorporating explicit visual cues like bounding boxes that require extra annotations. In addition, the resulting low-resolution crops often miss fine-grained details that MLLMs require for accurate reasoning. Therefore, we propose Mags-RL, an Agentic Reinforcement Learning (RL) framework that equips MLLMs with an external super-resolution"magnifying glass"agent for high-resolution fine-grained inspection. Specifically, the model performs two-round reasoning: in the first round, it generates an initial rationale and autonomously identifies regions of interest without relying on additional annotations; in the second round, it invokes a super-resolution agent to crop and upscale those regions, then revisits and verifies its earlier reasoning to produce the final answer. We also introduce a novel curriculum learning strategy that enables data-efficient RL training, needing as few as only 40 training samples to achieve reasonable performance. Experiments on VSR, TallyQA, and GQA subsets show its superior performance against recent strong competing methods, demonstrating high-quality reasoning with precise visual grounding. Code and weights will be released soon.
