news

May 31, 2026 The preprint GeoSVG-RL: Geometry-Aware Reinforcement Learning for Layout-Constrained Text-to-SVG Diagram Generation is available on arXiv.
Apr 01, 2026 I started as an assistant researcher in AI4Science at Shanghai AI Lab.
Jan 22, 2026 The preprint OptiSQL: Executable SQL Generation from Optical Tokens is available on arXiv.
Nov 07, 2025 SemVink: Advancing VLMs’ Semantic Understanding of Optical Illusions via Visual Global Thinking was presented as an oral paper at the EMNLP 2025 Main Conference in Suzhou.
Oct 14, 2025 The preprint Vision Language Models Map Logos to Text via Semantic Entanglement in the Visual Projector is available on arXiv.