Wenxuan Huang

Vision-R1. [ICLR2026] This is the first paper to explore how to effectively use R1-like RL for MLLMs and introduce Vision-R1, a reasoning MLLM that leverages cold-start initialization and RL training to incentivize reasoning capability.

Vision-DeepResearch. [ICML 2026] Multimodal deep-research MLLM and benchmark. The first long-horizon multimodal deep-research MLLM, extending the number of reasoning turns to dozens and the number of search-engine interactions to hundreds.

Awesome-Interleaving-Reasoning. Interleaving Reasoning: Next-Generation Reasoning Systems for AGI

Interleaving-Reasoning-Generation. [ICLR 2026] This is an early exploration to introduce Interleaving Reasoning to Text-to-image Generation field and achieve the SoTA benchmark performance. It also significantly improves the quality, fine-grained details and aesthetic aspects of generated images.

dynamic_llava. [ICLR 2025] The official pytorch implement of "Dynamic-LLaVA: Efficient Multimodal Large Language Models via Dynamic Vision-language Context Sparsification".

TokenExpansion. [CVPR 2024] The official pytorch implementation of "A General and Efficient Training for Transformer via Token Expansion".

Heartbeat-sequence-prediction. 天池上的一场长期赛（心跳信号分类预测），非常简单朴素的实现，长期赛榜单第8名（258.9817分）

KDFS. The official pytorch implementation of KDFS. Paper: Filter Pruning for Efficient CNNs via Knowledge-driven Differential Filter Sampler

Data-collection. 数据采集与预处理的实验

G-LA-MSG-and-AOG-PSPNet. Official pytorch implementation of "An Intelligent First Arrival Picking Method of Microseismic Signals Based on the Small Sample Expansion"

Tampered-Image-Detection. 天池真实场景篡改图像检测挑战赛

StyleGan-signal-pytorch. Modified StyleGan to be used to generate one-dimensional signals

Osilly-resume. VUE 搭建的个人简历

SE-Personal-programming. 软件工程的实验