University of Illinois at Urbana-Champaign · 61 records

dateidtitleauthorsvenue↑
2026-11-192611.24154Game2Player: Scaling Gameplay Experience for Visual Agents through Environment-Player Co-DevelopmentXirui Jiang et al.arXiv28
2026-11-192611.24059Replay, Don't Resample: Off-Policy Consolidation of Long-Horizon RL ExpertsZhenrui Yue et al.arXiv4
2026-11-182611.21791Regularized Score CenteringHao Bai et al.arXiv4
2026-11-172611.19649LoopFill: Pipelined Multi-Token Speculative Decoding for Looped TransformersSiqi Zhu et al.arXiv4
2026-11-102611.11994EmoHarness: Boosting Multimodal Emotion Understanding with Meta-Harness EvolvingFenghai Li et al.arXiv2
2026-10-252610.34073Memory Tax: Dissecting the Net Value of Agent MemoryXuan He et al.arXiv29
2026-10-212610.27450Ask to Distill: Learning to Synthesize Data for On-Policy DistillationJiashuo Sun et al.arXiv22
2026-10-212610.29927Your Benchmark Is an Empirical Measure Over DifficultyYifan Sun et al.arXiv18
2026-10-202610.22894Competence-Based Evaluation of Logical Reasoning in Large Language ModelsJize Jiang et al.arXiv4
2026-10-162610.21412JevAgentHarness: Benchmarking, Deploying and Training Decision Models That Share Control with LLMsKunlun Zhu et al.arXiv25
2026-10-142610.18738Enabling Token-wise Importance Sampling in Off-Policy RLVR with Proximal KL RegularizationHao Bai et al.arXiv3
2026-10-142610.04851CURIO: Curiosity-Driven Test-Time Learning for Open-Ended DiscoveryTao Feng et al.arXiv3
2026-10-092610.04977Do LLMs Understand Sequential Structure? A Controlled Study of Inference and GenerationJerry Wang et al.arXiv13
2026-10-082610.09457DSReg: Provably Recovering Individual World Latents without ReconstructionYujia Zheng et al.arXiv3
2026-10-082609.38680ReGain: Restoring Subject Fidelity in Personalization on Synthetic ImagesShubhang Bhatnagar et al.arXiv1
2026-10-072610.07510Understanding and Enhancing Backdoor Persistency in LLM Agent Post-TrainingQiusi Zhan et al.arXiv13
2026-10-062610.04292LMBuild: Evaluating LLM Agents for Generating Buildable and Functional StructuresJiateng Liu et al.arXiv34
2026-10-052609.38078MotorMind: Scaffolding General Vision Language Models for Zero-Shot Robot ManipulationBingxuan Li et al.arXiv55
2026-10-052610.02179From Gradients to Capabilities: Understanding Multi-Teacher On-Policy DistillationSiqi Zhu et al.arXiv11
2026-10-052609.39050Covert Assistance: Helpful LLM Agents Evade Oversight in Multi-Agent SystemsDeema Alnuhait et al.arXiv11
2026-10-022610.02193Hierarchical Continuous Diffusion Language ModelsHui Ren et al.arXiv74
2026-10-022610.02196InterEvolve: Test-Time Evolution of Reward Programs for Humanoid Loco-ManipulationZhuo Lin et al.arXiv31
2026-09-302609.33326ANTMAN: Adaptive Need Tracking for Multi-Agent Navigation in Large Information SpacesJerry Wang et al.arXiv23
2026-09-302609.33591Pretraining Transformers with Quantized Softmax in AttentionShangzhen Zhu et al.arXiv6
2026-09-292609.33586Approximating Softmax in Pretrained LLMs: Model Sensitivity and Kernel AccelerationShangzhen Zhu et al.arXiv4
2026-09-292609.36219LeRF: Learning Reference Coordinate Frames for Perspective Taking ReasoningBang Xiao et al.arXiv2
2026-09-142609.12101Competence-Gated Pooling of Language Models and Priors for Event ForecastingAditi Tiwari et al.arXiv2
2026-09-142609.12078Feature Recovery for Object Understanding After Irreversible Fire DamageAditi Tiwari et al.arXiv1
2026-09-102608.12564Scaling Automatic Research Agents via World ModelsXiyuan Yang et al.arXiv82
2026-09-012608.28833Evaluating the Hidden Costs of Personalization in Large Language ModelsYumeng Wang et al.arXiv30
2026-08-272608.19583VGI-Bench: Probing Visual Intelligence in Video Generation ModelsXuan He et al.arXiv57
2026-08-242608.21278CLEAR: Continuous Latent Adapter Routing for Utility-Preserving LLM Safety AlignmentChengxiao Wang et al.arXiv6
2026-08-142608.06867LLMRouter: Unified Infrastructure for Developing, Evaluating, and Deploying LLM RoutersTao Feng et al.arXiv49
2026-08-132608.12307AI4AI at Test-Time: Strong-to-Weak Capability Transfer via HarnessesCheng Qian et al.arXiv107
2026-08-102607.26178DuplexGen: Adaptive Synthesis of Human-AI Turn-Taking DialoguesTakyoung Kim et al.arXiv11
2026-07-312607.27372Explorative Modeling: Unlocking a Third Pretraining Axis and End-to-End GenerationAlexi Gladstone et al.arXiv17
2026-07-222607.18754AgentDebugX: An Open-Source Toolkit for Failure Observability, Attribution, and Recovery in LLM AgentsKunlun Zhu et al.arXiv24
2026-07-202607.10966SVR-R1: Bootstrapping Multi-modal Reasoning with Self-verification in Reinforcement LearningMingyuan Wu et al.arXiv4
2026-07-072607.00597Multi-Turn Agentic Scientific Literature Search via Workflow InductionJisen Li et al.arXiv23
2026-06-302606.24256Trimming the Long-Tail of Visual World Modeling EvaluationBingxuan Li et al.arXiv31
2026-06-292606.28187GBC: Gradient-Based Connections for Optimizing Multi-Agent SystemsXiaocheng Yang et al.arXiv13
2026-06-232606.22388PlanBench-XL: Evaluating Long-Horizon Planning of LLM Tool-Use Agents in Large-Scale Tool EcosystemsJiayu Liu et al.arXiv92
2026-06-172606.15134Beyond Scalar Distances: Semantic Attribute Gradients from Frozen MLLMs for Visual EmbeddingsShubhang Bhatnagar et al.arXiv2
2026-06-162606.15956You Don't Need Strong Assumptions: Visual Representation Learning via Temporal DifferencesNinad Daithankar et al.arXiv12
2026-06-112606.11482Building Social World Models with Large Language ModelsHaofei Yu et al.arXiv4
2026-06-102606.10061BenSyc: Benchmarking Conversational Sycophancy and Human Alignment in LLMs for Bengali ContextsKazi Noshin et al.arXiv1
2026-06-052606.05622AdaPlanBench: Evaluating Adaptive Planning in Large Language Model Agents under World and User ConstraintsJiayu Liu et al.arXiv44
2026-05-292605.26029CausaLab: A Scalable Environment for Interactive Causal Discovery Toward AI ScientistsJunlin Yang et al.arXiv13
2026-05-282605.26396Advancing Creative Physical Intelligence in Large Multimodal ModelsCheng Qian et al.arXiv19
2026-05-222605.22642Spreadsheet-RL: Advancing Large Language Model Agents on Realistic Spreadsheet Tasks via Reinforcement LearningBanghao Chi et al.arXiv35
2026-05-202605.15514RoPE Distinguishes Neither Positions Nor Tokens in Long Contexts, ProvablyYufeng Du et al.arXiv
2026-05-192605.15764GRASP: Learning to Ground Social Reasoning in Multi-Person Non-Verbal InteractionsJunho Kim et al.arXiv3
2026-05-152605.00180RouteProfile: Elucidating the Design Space of LLM Profiles for RoutingJingjun Xu et al.arXiv29
2026-05-152605.15012Boosting Reinforcement Learning with Verifiable Rewards via Randomly Selected Few-Shot GuidanceKai Yan et al.arXiv2
2026-05-142605.12975Retrieval is Cheap, Show Me the Code: Executable Multi-Hop Reasoning for Retrieval-Augmented GenerationJiashuo Sun et al.arXiv8
2026-05-132605.12978Useful Memories Become Faulty When Continuously Updated by LLMsDylan Zhang et al.arXiv18
2026-05-132605.11182The Many Faces of On-Policy Distillation: Pitfalls, Mechanisms, and FixesSiqi Zhu et al.arXiv4
2026-05-112605.05558Who Prices Cognitive Labor in the Age of Agents? Compute-Anchored WagesSiqi ZhuarXiv1
2026-05-072605.02910CreativityBench: Evaluating Agent Creative Reasoning via Affordance-Based Tool RepurposingCheng Qian et al.arXiv22
2026-05-052605.01214Agentic AI Systems Should Be Designed as Marginal Token AllocatorsSiqi ZhuarXiv2
2026-05-012604.27351Heterogeneous Scientific Foundation Model CollaborationZihao Li et al.arXiv127