rewardmap-sparse-rewards-visual-reasoning

Improve multimodal LLMs on fine-grained visual reasoning tasks (e.g., reading transit maps) by decomposing training into stages: basic perception (VQA) -> simple reasoning -> complex spatial reasoning. Incorporates 'detail rewards' for intermediate visual understanding, bootstrapping models from simple to complex tasks while addressing sparse reward challen…

Independent GitHub snapshot · Automated classification · Original documentation governs