17 papers
Words to World: Benchmarking LLMs in Interactive Physical Environments
Xinnan Dai, Kaiqi Yang, Yaxin Li +9
Words are increasingly becoming an interface for action: large language models (LLMs) can now invoke tools, operate devices, and make decisions that affect the physical world. Howe…
Universal Test-Time Training
Zefan Cai, Qinzhe Hu, Ziqiao Ma +2
Recent Test-Time Training (TTT) architectures compress context into fast weights that are updated online and queried as memory. Existing TTT designs keep this memory private to eac…
Multi-Head Attention Residuals
Cheng Luo, Zefan Cai, Junjie Hu
Transformers propagate information across depth through a single additive residual stream: every sublayer reads only the most recent state. Attention residuals relax this by lettin…
Chimera: Designing and Chinchilla-Scaling Hybrid Visual Diffusion Transformers
Chongjian Ge, Hanwen Jiang, Tianyu Wang +9
Visual generation increasingly requires high-resolution images, long videos, and multimodal context, making the quadratic cost of full attention prohibitive. We introduce Chimera,…
UniTemp: Unlocking Video Generation in Any Temporal Order via Bidirectional Distillation
Lin Zhang, Sicheng Mo, Zefan Cai +6
Autoregressive video diffusion models have emerged as a promising approach for long video generation, achieving strong performance in streaming settings. However, existing methods…
Test-Time Training with Next-Token Prediction
Xuan Ouyang, Zefan Cai, Junjie Hu
Next-token prediction is the self-supervised signal that trains language models, and every observed prompt token provides the same signal at test time. We study whether this signal…
Delta Attention Residuals
Cheng Luo, Zefan Cai, Junjie Hu
Attention Residuals replace standard additive residual connections with learned softmax attention over previous layer outputs, enabling selective cross-layer routing. However, stan…
From Preferences to Prejudice: The Role of Alignment Tuning in Shaping Social Bias in Video Diffusion Models
Zefan Cai, Haoyi Qiu, Haozhe Zhao +6
Recent advances in video diffusion models have significantly enhanced text-to-video generation, particularly through alignment tuning using reward models trained on human preferenc…
Scaling Up Audio-Synchronized Visual Animation: An Efficient Training Paradigm
Lin Zhang, Zefan Cai, Yufan Zhou +10
Recent advances in audio-synchronized visual animation enable control of video content using audios from specific classes. However, existing methods rely heavily on expensive manua…
CoLD: Counterfactually-Guided Length Debiasing for Process Reward Models in Mathematical Reasoning
Congmin Zheng, Jiachen Zhu, Jianghao Lin +6
Process Reward Models (PRMs) play a central role in evaluating and guiding multi-step reasoning in large language models (LLMs), especially for mathematical problem solving. Howeve…
MENTOR: Efficient Multimodal-Conditioned Tuning for Autoregressive Vision Generation Models
Haozhe Zhao, Zefan Cai, Shuzheng Si +5
Recent text-to-image models produce high-quality results but still struggle with precise visual control, balancing multimodal inputs, and requiring extensive training for complex m…
R-KV: Redundancy-aware KV Cache Compression for Reasoning Models
Zefan Cai, Wen Xiao, Hanshi Sun +11
Reasoning models have demonstrated impressive performance in self-reflection and chain-of-thought reasoning. However, they often produce excessively long outputs, leading to prohib…
VisualToolAgent (VisTA): A Reinforcement Learning Framework for Visual Tool Selection
Zeyi Huang, Yuyang Ji, Anirudh Sundara Rajan +5
We introduce VisTA, a new reinforcement learning framework that empowers visual agents to dynamically explore, select, and combine tools from a diverse library based on empirical p…
HeadInfer: Memory-Efficient LLM Inference by Head-wise Offloading
Cheng Luo, Zefan Cai, Hanshi Sun +7
Transformer-based large language models (LLMs) demonstrate impressive performance in long context generation. Extending the context length has disproportionately shifted the memory…
AdaptiveStep: Automatically Dividing Reasoning Step through Model Confidence
Yuliang Liu, Junjie Lu, Zhaoling Chen +10
Current approaches for training Process Reward Models (PRMs) often involve breaking down responses into multiple reasoning steps using rule-based techniques, such as using predefin…
No Preference Left Behind: Group Distributional Preference Optimization
Binwei Yao, Zefan Cai, Yun-Shiuan Chuang +4
Preferences within a group of people are not uniform but follow a distribution. While existing alignment methods like Direct Preference Optimization (DPO) attempt to steer models t…
COMMA: A Communicative Multimodal Multi-Agent Benchmark
Timothy Ossowski, Danyal Maqbool, Jixuan Chen +3
The rapid advances of multimodal agents built on large foundation models have largely overlooked their potential for language-based communication between agents in collaborative ta…