← Search

Peter Henry Tu

2 accepted papers

2026

More Thought, Less Accuracy? On the Dual Nature of Reasoning in Vision-Language Models

ICLR 2026poster

Reasoning has emerged as a pivotal capability in Large Language Models (LLMs). Through Reinforcement Learning (RL), typically Group Relative Policy Optimization (GRPO), these models are able to solve complex tasks such as mathematics and code generation. Building on these advances, recent research h…

Cited by 0SourcecodeScholar
2025

Probability Density Geodesics in Image Diffusion Latent Space

CVPR 2025poster

Diffusion models indirectly estimate the probability density over a data space, which can be used to study its structure. In this work, we show that geodesics can be computed in diffusion latent space, where the norm induced by the spatially-varying inner product is inversely proportional to the pro…