13 papers
Frustratingly Easy Majority Voting Using Length Conditioning
Fiorenzo Parascandolo, Enver Sangineto, Xiaoyi Yu +4
Majority Voting (MV) is a simple and widely used strategy which improves the reasoning accuracy of Large Language Models (LLMs) by allocating additional inference-time computation.…
ROMA: LLM System for Real-World Object-Centric Multi-Sensory Active Perception
Ruoxuan Feng, Yutong Chen, Ruihua Song +4
Humans inherently understand the physical world through an active process. When sensory evidence is insufficient to infer physical properties, we naturally interact with the enviro…
Can Computation from Earlier Problems Help LLMs Solve New Ones?
Jipei He, Wenhui Tan, Xiaoyi Yu +4
Large language models often solve independent problems in the same conversation. Can computation from earlier problems help them solve new ones? To answer this question, we first c…
Informed Masking: Structure-Aware Perturbation for Reinforcement Learning in Diffusion Large Language Models
Xiaoyi Yu, Enver Sangineto, Pei Fu +6
Diffusion Large Language Models (dLLMs) have emerged as an efficient alternative to autoregressive models, yet aligning them via Reinforcement Learning (RL) requires likelihood sur…
Pair-In, Pair-Out: Latent Multi-Token Prediction for Efficient LLMs
Wenhui Tan, Minghao Li, Xiaoqian Ma +5
Long chain-of-thought reasoning has made autoregressive decoding the dominant inference cost of modern large language models. Existing methods target either the input side (latent…
MSJoE: Jointly Evolving MLLM and Sampler for Efficient Long-Form Video Understanding
Wenhui Tan, Xiaoyi Yu, Jiaze Li +5
Efficiently understanding long-form videos remains a fundamental challenge for multimodal large language models (MLLMs). In this paper, we present MLLM-Sampler Joint Evolution (MSJ…
BFS-PO: Best-First Search for Large Reasoning Models
Fiorenzo Parascandolo, Wenhui Tan, Enver Sangineto +2
Large Reasoning Models (LRMs) such as OpenAI o1 and DeepSeek-R1 have shown excellent performance in reasoning tasks using long reasoning chains. However, this has also led to a sig…
Restoring Exploration after Post-Training: Latent Exploration Decoding for Large Reasoning Models
Wenhui Tan, Fiorenzo Parascandolo, Enver Sangineto +6
Large Reasoning Models (LRMs) have recently achieved strong mathematical and code reasoning performance through Reinforcement Learning (RL) post-training. However, we show that mod…
Think-Clip-Sample: Slow-Fast Frame Selection for Video Understanding
Wenhui Tan, Ruihua Song, Jiaze Li +2
Recent progress in multi-modal large language models (MLLMs) has significantly advanced video understanding. However, their performance on long-form videos remains limited by compu…
PlanMoGPT: Flow-Enhanced Progressive Planning for Text to Motion Synthesis
Chuhao Jin, Haosen Li, Bingzi Zhang +7
Recent advances in large language models (LLMs) have enabled breakthroughs in many multimodal generation tasks, but a significant performance gap still exists in text-to-motion gen…
Think Silently, Think Fast: Dynamic Latent Compression of LLM Reasoning Chains
Wenhui Tan, Jiaze Li, Jianzhong Ju +3
Large Language Models (LLMs) achieve superior performance through Chain-of-Thought (CoT) reasoning, but these token-level reasoning chains are computationally expensive and ineffic…
Think-Then-React: Towards Unconstrained Human Action-to-Reaction Generation
Wenhui Tan, Boyuan Li, Chuhao Jin +3
Modeling human-like action-to-reaction generation has significant real-world applications, like human-robot interaction and games. Despite recent advancements in single-person moti…
RoLD: Robot Latent Diffusion for Multi-task Policy Modeling
Wenhui Tan, Bei Liu, Junbo Zhang +2
Modeling generalized robot control policies poses ongoing challenges for language-guided robot manipulation tasks. Existing methods often struggle to efficiently utilize cross-data…