23 papers
IQuest-Q1: Advancing Foundation Capabilities for Agentic CLI Systems
A. Jian Yang, An Huang, Bo Dan +173
We present IQuest-Q1, a sparse Mixture-of-Experts (MoE) agentic foundation model with 320B total parameters (15B activated). IQuest-Q1 is developed through a three-stage process: (…
Video2World: Benchmarking Coding Agents for Interactive World Modeling from Embodied Videos
Jinzhou Tang, Zijun Zhang, Jing Yang +10
Building interactive simulators from real-world observations is a promising way to scale embodied data, but current pipelines still rely heavily on manual environment construction…
Online Video Agent Harness for Long Video Understanding
Sen Yang, Boqiang Duan, Jing Yang +7
Long video understanding often behaves like a visual needle-in-a-haystack problem: query-relevant evidence is sparsely distributed across long temporal spans, while packing dense f…
Spectral Gating Networks
Jusheng Zhang, Yijia Fan, Kaitong Cai +5
Gating mechanisms are ubiquitous, yet a complementary question in feed-forward networks remains under-explored: how to introduce frequency-rich expressivity without sacrificing sta…
Rational ANOVA Networks
Jusheng Zhang, Ningyuan Liu, Qinhan Lyu +2
Deep neural networks typically treat nonlinearities as fixed primitives (e.g., ReLU), limiting both interpretability and the granularity of control over the induced function class.…
Why Keep Your Doubts to Yourself? Trading Visual Uncertainties in Multi-Agent Bandit Systems
Jusheng Zhang, Yijia Fan, Kaitong Cai +6
Vision-Language Models (VLMs) enable powerful multi-agent systems, but scaling them is economically unsustainable: coordinating heterogeneous agents under information asymmetry oft…
Self-Rewarded Multimodal Coherent Reasoning Across Diverse Visual Domains
Jesen Zhang, Ningyuan Liu, Kaitong Cai +5
Multimodal LLMs often produce fluent yet unreliable reasoning, exhibiting weak step-to-step coherence and insufficient visual grounding, largely because existing alignment approach…
RevFFN: Memory-Efficient Full-Parameter Fine-Tuning of Mixture-of-Experts LLMs with Reversible Blocks
Ningyuan Liu, Jing Yang, Kaitong Cai +1
Full parameter fine tuning is a key technique for adapting large language models (LLMs) to downstream tasks, but it incurs substantial memory overhead due to the need to cache exte…
FlashVLM: Text-Guided Visual Token Selection for Large Multimodal Models
Kaitong Cai, Jusheng Zhang, Jing Yang +4
Large vision-language models (VLMs) typically process hundreds or thousands of visual tokens per image or video frame, incurring quadratic attention cost and substantial redundancy…
SirenPose: Dynamic Scene Reconstruction via Geometric Supervision
Kaitong Cai, Jensen Zhang, Jing Yang +1
We introduce SirenPose, a geometry-aware loss formulation that integrates the periodic activation properties of sinusoidal representation networks with keypoint-based geometric sup…
Reflective Confidence: Correcting Reasoning Flaws via Online Self-Correction
Qinglin Zeng, Jing Yang, Keze Wang
Large language models (LLMs) have achieved strong performance on complex reasoning tasks using techniques such as chain-of-thought and self-consistency. However, ensemble-based app…
Large Language Models as Discounted Bayesian Filters
Jensen Zhang, Jing Yang, Keze Wang
Large Language Models (LLMs) demonstrate strong few-shot generalization through in-context learning, yet their reasoning in dynamic and stochastic environments remains opaque. Prio…
MM-CoT:A Benchmark for Probing Visual Chain-of-Thought Reasoning in Multimodal Models
Jusheng Zhang, Kaitong Cai, Xiaoyang Guo +10
The ability to perform Chain-of-Thought (CoT) reasoning marks a major milestone for multimodal models (MMs), enabling them to solve complex visual reasoning problems. Yet a critica…
Causal Invariance and Counterfactual Learning Driven Cooperative Game for Multi-Label Classification
Yijia Fan, Jusheng Zhang, Kaitong Cai +2
Multi-label classification (MLC) remains vulnerable to label imbalance, spurious correlations, and distribution shifts, challenges that are particularly detrimental to rare label p…
3DAlign-DAER: Dynamic Attention Policy and Efficient Retrieval Strategy for Fine-grained 3D-Text Alignment at Scale
Yijia Fan, Jusheng Zhang, Kaitong Cai +3
Despite recent advancements in 3D-text cross-modal alignment, existing state-of-the-art methods still struggle to align fine-grained textual semantics with detailed geometric struc…
Cost-Effective Communication: An Auction-based Method for Language Agent Interaction
Yijia Fan, Jusheng Zhang, Kaitong Cai +4
Multi-agent systems (MAS) built on large language models (LLMs) often suffer from inefficient "free-for-all" communication, leading to exponential token costs and low signal-to-noi…
Guardian: Decoupling Exploration from Safety in Reinforcement Learning
Kaitong Cai, Jusheng Zhang, Jing Yang +1
Hybrid offline--online reinforcement learning (O2O RL) promises both sample efficiency and robust exploration, but suffers from instability due to distribution shift between offlin…
RaCoT: Plug-and-Play Contrastive Example Generation Mechanism for Enhanced LLM Reasoning Reliability
Kaitong Cai, Jusheng Zhang, Yijia Fan +2
Retrieval-Augmented Generation (RAG) faces a core bottleneck with knowledge-sparse and semantically ambiguous long-tail queries, where retrieval noise distorts reasoning and necess…
Agent-GSPO: Communication-Efficient Multi-Agent Systems via Group Sequence Policy Optimization
Yijia Fan, Jusheng Zhang, Jing Yang +1
To combat the prohibitive communication costs of ``free-for-all" multi-agent systems (MAS), we introduce \textbf{Agent-GSPO}, a framework that directly optimizes for token economy…
Backward-Friendly Optimization: Training Large Language Models with Approximate Gradients under Memory Constraints
Jing Yang, Kaitong Cai, Yijia Fan +2
Full fine-tuning of Large Language Models (LLMs) is notoriously memory-intensive, primarily because conventional optimizers such as SGD or Adam assume access to exact gradients der…
Learning Dynamics of VLM Finetuning
Jusheng Zhang, Kaitong Cai, Jing Yang +1
Preference-based finetuning of vision--language models (VLMs) is brittle: trivially wrong negatives inject uninformative gradients that destabilize training. We recast alignment as…
Top-Down Semantic Refinement for Image Captioning
Jusheng Zhang, Kaitong Cai, Jing Yang +3
Large Vision-Language Models (VLMs) face an inherent contradiction in image captioning: their powerful single-step generation capabilities often lead to a myopic decision-making pr…
HiVA: Self-organized Hierarchical Variable Agent via Goal-driven Semantic-Topological Evolution
Jinzhou Tang, Jusheng Zhang, Qinhan Lv +4
Autonomous agents play a crucial role in advancing Artificial General Intelligence, enabling problem decomposition and tool orchestration through Large Language Models (LLMs). Howe…