14 papers
4DMemory: Benchmarking Long-Horizon Spatiotemporal Memory with Harness-Augmented Agents
Yanhao Jia, Yi Zhu, Yiran Zhong +7
Humans efficiently process multimodal streams spanning hours to years, building persistent, fine-grained spatiotemporal memories that support reasoning about and predicting long-ho…
Qwen-Planner-Agent: A Closed-Loop AI-for-AI Framework for Real-World Mobile Planner Agents
Tingyu Qu, Weigao Sun, Yuecheng Liu +23
The rapid progression of large language models is extending AI from passive content generation into the active workflows of engineering and scientific discovery. This shift raises…
ProCredit: From Outcome Rewards to Progress Credit in Agentic Reinforcement Learning
Ming Ma, Yi Zhu, Yiran Zhong +7
Long-horizon agentic tasks require an agent to modify an environment through a sequence of tool calls, with success determined by the final state. The standard recipe assigns a sin…
Block-Sparse Attention with Semantic-Geometric Decoupled Routing
Xinwei Long, Weigao Sun, Weibo Gao +6
Long-context inference has become a defining capability of large language models, but exact dense attention remains costly due to its quadratic scaling with sequence length. Block-…
ReGraph: Learning to Generate Recipe Graphs from Food Images
Guoshan Liu, Bin Zhu, Pengkun Jiao +3
Recent Large Multimodal Models (LMMs) have achieved impressive performance in recipe generation from food images.However, cooking is a structured transformation process in which in…
Disentangling Semantic Attention from Structural Bias in the Attention Manifold
Pengkun Jiao, Bin Zhu, Jingjing Chen +1
The empirical success of attention mechanism in Multimodal Large Language Models (MLLMs) often obscures its inherent, subtle flaws. Specifically, MLLMs consistently exhibit disprop…
Spatiotemporal Sycophancy: Negation-Based Gaslighting in Video Large Language Models
Ziyao Tang, Pengkun Jiao, Bin Zhu +3
Video Large Language Models (Vid-LLMs) have demonstrated remarkable performance in video understanding tasks, yet their robustness under conversational interaction remains largely…
Predicting Future Utility: Global Combinatorial Optimization for Task-Agnostic KV Cache Eviction
Ziyao Tang, Pengkun Jiao, Xinhang Chen +3
Given the quadratic complexity of attention, KV cache eviction is vital to accelerate model inference. Current KV cache eviction methods typically rely on instantaneous heuristic m…
Dual-LoRA and Quality-Enhanced Pseudo Replay for Multimodal Continual Food Learning
Xinlan Wu, Bin Zhu, Feng Han +2
Food analysis has become increasingly critical for health-related tasks such as personalized nutrition and chronic disease prevention. However, existing large multimodal models (LM…
Don't Deceive Me: Mitigating Gaslighting through Attention Reallocation in LMMs
Pengkun Jiao, Bin Zhu, Jingjing Chen +2
Large Multimodal Models (LMMs) have demonstrated remarkable capabilities across a wide range of tasks. However, their vulnerability to user gaslighting-the deliberate use of mislea…
Domain Expansion and Boundary Growth for Open-Set Single-Source Domain Generalization
Pengkun Jiao, Na Zhao, Jingjing Chen +1
Open-set single-source domain generalization aims to use a single-source domain to learn a robust model that can be generalized to unknown target domains with both domain shifts an…
From Holistic to Localized: Local Enhanced Adapters for Efficient Visual Instruction Fine-Tuning
Pengkun Jiao, Bin Zhu, Jingjing Chen +2
Efficient Visual Instruction Fine-Tuning (EVIT) seeks to adapt Multimodal Large Language Models (MLLMs) to downstream tasks with minimal computational overhead. However, as task di…
Unlocking Textual and Visual Wisdom: Open-Vocabulary 3D Object Detection Enhanced by Comprehensive Guidance from Text and Image
Pengkun Jiao, Na Zhao, Jingjing Chen +1
Open-vocabulary 3D object detection (OV-3DDet) aims to localize and recognize both seen and previously unseen object categories within any new 3D scene. While language and vision f…
RoDE: Linear Rectified Mixture of Diverse Experts for Food Large Multi-Modal Models
Pengkun Jiao, Xinlan Wu, Bin Zhu +3
Large Multi-modal Models (LMMs) have significantly advanced a variety of vision-language tasks. The scalability and availability of high-quality training data play a pivotal role i…