Yang Minghao — accepted AI-conference papers

2026

AG-REPA: Causal Layer Selection for Representation Alignment in Audio Flow Matching

Pengfei Zhang, Tianxin Xie, Yang Minghao, Li Liu

ICML 2026poster

REPresentation Alignment (REPA) improves the training of generative flow models by aligning intermediate hidden states with pretrained teacher features, but its effectiveness in token-conditioned audio Flow Matching critically depends on the choice of supervised layers, which is typically made heuri…

Cited by 0 Source ↗Scholar ↗

2025

VaporTok: RL-Driven Adaptive Video Tokenizer with Prior & Task Awareness

Yang Minghao, Zechen Bai, Jing Lin, Haoqian Wang, Alex Jinpeng Wang

NeurIPS 2025poster

Recent advances in visual tokenizers have demonstrated their effectiveness for multimodal large language models and autoregressive generative models. However, most existing visual tokenizers rely on a fixed downsampling rate at a given visual resolution, and consequently produce a constant number of…

Cited by 0 Source ↗Scholar ↗