7 papers
SoftmaxGRPO: Learning to Reason using Softmax Advantage Group Estimation
Jefferson Hernandez, Jaywon Koo, Zilin Xiao +2
Group-based reinforcement learning objectives such as GRPO can allocate learning signal poorly across prompt difficulty: under binary rewards, group normalization induces a diverge…
Agentic Discovery with Active Hypothesis Exploration for Visual Recognition
Jaywon Koo, Jefferson Hernandez, Ruozhen He +3
We introduce HypoExplore, an agentic framework that formulates neural architecture discovery for visual recognition as a hypothesis-driven scientific inquiry. Given a human-specifi…
Text-to-Stage: Spatial Layouts from Long-form Narratives
Jefferson Hernandez, Swarnadeep Saha, Chenxi Whitehouse +6
In this work, we probe the ability of a language model to demonstrate spatial reasoning from unstructured text, mimicking human capabilities and automating a process that benefits…
ProxyThinker: Test-Time Guidance through Small Visual Reasoners
Zilin Xiao, Jaywon Koo, Siru Ouyang +3
Recent advancements in reinforcement learning with verifiable rewards have pushed the boundaries of the visual reasoning capabilities in large vision-language models (LVLMs). Howev…
Improving Large Vision and Language Models by Learning from a Panel of Peers
Jefferson Hernandez, Jing Shi, Simon Jenni +2
Traditional alignment methods for Large Vision and Language Models (LVLMs) primarily rely on human-curated preference data. Human-generated preference data is costly; machine-gener…
Evaluating Text-to-Image and Text-to-Video Synthesis with a Conditional Fréchet Distance
Jaywon Koo, Jefferson Hernandez, Moayed Haji-Ali +2
Evaluating text-to-image and text-to-video models is challenging due to a fundamental disconnect: established metrics fail to jointly measure visual quality and semantic alignment…