4 papers
MAD-Guard: Controlled Study of Autoregressive Generation versus Direct Decision Interfaces for Closed Multimodal Forensic Tasks
Hao Chen
When should multimodal foundation models generate tokens, and when should they directly output a decision? We present MAD-Guard, a controlled study of output-decision interfaces fo…
BitDance: Scaling Autoregressive Generative Models with Binary Tokens
Yuang Ai, Jiaming Han, Shaobin Zhuang +8
We present BitDance, a scalable autoregressive (AR) image generator that predicts binary visual tokens instead of codebook indices. With high-entropy binary latents, BitDance lets…
UniWeTok: An Unified Binary Tokenizer with Codebook Size for Unified Multimodal Large Language Model
Shaobin Zhuang, Yuang Ai, Jiaming Han +12
Unified Multimodal Large Language Models (MLLMs) require a visual representation that simultaneously supports high-fidelity reconstruction, complex semantic extraction, and generat…
The Image as Its Own Reward: Reinforcement Learning with Adversarial Reward for Image Generation
Weijia Mao, Hao Chen, Zhenheng Yang +1
A reliable reward function is essential for reinforcement learning (RL) in image generation. Most current RL approaches depend on pre-trained preference models that output scalar r…