Bring your perspective.
Read the discussion without an account. Sign in with GitHub to ask questions, reply and react. Your GitHub name and profile appear with your comments.
- 01Open GitHubUse an existing account or create one free.
- 02Choose a paperStart with one claim, equation, or result.
- 03Share what you learnInclude your model, baseline, and test conditions.
Read public conversations and follow the ones you care about.
Choose a paper
Start with a question about a claim, an equation, or what it would take to try the method.
1406 papers
agent evaluation Published 2026-10-15
FDE-Gym: The Last Mile of AI Delivery
arXiv 2610.22350efficient training Published 2026-10-15
MLExpert: Towards a Generalist Agentic Model for Autonomous Machine Learning Engineering
arXiv 2610.22344coding agents Published 2026-10-15
3DCodeVerse: Exploring 3D Worlds as Code
arXiv 2610.22341safety and guardrails Published 2026-10-15
Evaluation Awareness Beyond English: Do LLMs Act Differently When Tested in Indic Languages?
arXiv 2610.22333efficient training Published 2026-10-15
Training Reasoning Models to Generate Diverse Scientific Ideas
arXiv 2610.22324multi-agent orchestration Published 2026-10-15
Learning to Steer LLM Agents and Multi-Agent Systems
arXiv 2610.22316multi-agent orchestration Published 2026-10-15
LEASH: Uncertainty-Aware Weak-to-Strong Escalation and Oversight for Computer-Use Agents
arXiv 2610.22280memory and context Published 2026-10-15
Librarian Agents: Training LLMs to Curate Long-Context Workspaces that Anticipate Future Queries
arXiv 2610.22273safety and guardrails Published 2026-10-15
Language Is a Security Surface for Agent Guards
arXiv 2610.22244tool use Published 2026-10-15
OS-Helix: Unifying GUI Interaction and Code Execution in Native Computer-Using Agents
arXiv 2610.22234agent evaluation Published 2026-10-15
Beyond Downstream QA: Rethinking How We Evaluate Agent Memory
arXiv 2610.22209multi-agent orchestration Published 2026-10-15
AutoSFT: An Agentic System for Automated Data Research in Supervised Fine-Tuning
arXiv 2610.22194serving and batching Published 2026-10-15
Efficient KV Cache Migration for Geo-Distributed LLM Inference in Collaborative Edge Computing
arXiv 2610.22172agent evaluation Published 2026-10-15
PROBE: Learning to Audit Policy Compliance in Tool-Using LLM Agents
arXiv 2610.22124efficient training Published 2026-10-15
Unfragmenting Nawatl NMT: Single-Variant Specialization for Masewalkopa
arXiv 2610.22110agent evaluation Published 2026-10-15
WEAVE: Weaving Real-World Code Evolution into Dependency-Structured Tasks for Benchmarking and Training Code Agents
arXiv 2610.22039agent evaluation Published 2026-10-15
GraphACM: Graph-based Action Consistency Measurement for Uncertainty Quantification in LLM Agents
arXiv 2610.22031multi-agent orchestration Published 2026-10-15
Augment the Cache, Replace the Activation: A 2x2 Study of Internal-State Sharing
arXiv 2610.22026coding agents Published 2026-10-15
Priming Base Models for Agentic Post-Training: An Open Midtraining Recipe
arXiv 2610.22019KV cache Published 2026-10-15
Close Enough is Nowhere Near: Fidelity-Consumability Decoupling in KV Caches
arXiv 2610.22016other Published 2026-10-15
GraphBEAT: Graph-based Bucketized Event Aggregation Over Time for Early Risk Stratification of Gastrointestinal Cancers
arXiv 2610.22014coding agents Published 2026-10-15
Benchmarking and Enhancing Design-Aware Code Review in the Linux Kernel
arXiv 2610.21925memory and context Published 2026-10-15
TEM: Temporal State Tracking Memory for Large Language Model Agents
arXiv 2610.21900quantization Published 2026-10-15
Compiling Multi-Threshold Operators for Quantized Non-Monotonic and Fine-Granular Activations in Custom-Tailored DNN Accelerators
arXiv 2610.21825quantization Published 2026-10-15
N-Gram Memories at Two Bits: Quantizing Lookup Tables in Large Language Models
arXiv 2610.21810quantization Published 2026-10-15
RoleQ: Role-Aligned Scales for Converting Language Models to Ternary
arXiv 2610.21808multi-agent orchestration Published 2026-10-15
RSIFlow: Adaptive Self-Improvement Workflows with an Evolving Meta-Agent
arXiv 2610.21791planning and decomposition Published 2026-10-15
DeciCAD: Harness-Grounded Learning of Action Policies for CAD Agents
arXiv 2610.21755multi-agent orchestration Published 2026-10-15
The Delegation Dilemma: Economies of Scale, Accountability, and Polycentric Governance
arXiv 2610.21746sparsity and pruning Published 2026-10-15
Sparse Federated Echo State Networks for Time-Series Classification via Smooth-L1 Regularization
arXiv 2610.21742failure recovery Published 2026-10-15
Human Interventions as Labels: Localizing, Steering, and (Sometimes) Editing the Computations Behind LLM Agent Failures
arXiv 2610.21708quantization Published 2026-10-15
ShiftQAT: Quantization-Aware Training for the Numerical Contract of a Fixed-Function Edge CNN Accelerator
arXiv 2610.21695agent evaluation Published 2026-10-15
AgentPerturb: Evaluating LLM Agent Robustness under Controlled Interaction Perturbations
arXiv 2610.21624speculative decoding Published 2026-10-15
LoopAhead: Speculating Ahead for Faster Decoding in Looped Language Models
arXiv 2610.21594memory and context Published 2026-10-15
Rethinking the Role of Tool Graph Memory in Multi-Turn Tool Use
arXiv 2610.21571agent evaluation Published 2026-10-15
Independent Data-Integrity Verification for LLM-Driven Functional Coverage Closure in Hardware Design
arXiv 2610.21570agent evaluation Published 2026-10-15
DialFork: Going Beyond Factuality to Evaluate Procedural Hallucination in Conversational Agents
arXiv 2610.21561other Published 2026-10-15
Reducing the Cost of Qudit State Preparation by Learning a Sparse Structure for SNAP Gates
arXiv 2610.21557multi-agent orchestration Published 2026-10-15
Agent Status Protocol (ASP): A Lightweight Specification for State Reporting in Multi-Agent Systems
arXiv 2610.21538KV cache Published 2026-10-15
When the KV Cache Is Slower Than the GPU: Scaling Remote KV Tiers for Context-Parallel Prefill
arXiv 2610.21500speculative decoding Published 2026-10-15
Trajectory-Level Speculation with Semantic Verification for Tool-Using Agents
arXiv 2610.21463tool use Published 2026-10-15
Incentivizing Tool-Use Reinforcement Learning for Post-Arrival Emergency Care Prediction
arXiv 2610.21449speculative decoding Published 2026-10-15
HyCo: Hybrid Drafting for Cost-Efficient Mobile-Cloud Speculative Decoding
arXiv 2610.21436tool use Published 2026-10-15
JevAgentHarness: Benchmarking, Deploying and Training Decision Models That Share Control with LLMs
arXiv 2610.21412memory and context Published 2026-10-15
MemCut: Every Secret Leaves a Trail in the Long-Term Memory of LLM Agents
arXiv 2610.21405other Published 2026-10-15
Fault Detection and Bit-Error Rate Diagnostics in Optical Wireless Links Using Edge-AI
arXiv 2610.21402sparsity and pruning Published 2026-10-15
Calibration Language Shapes Block-Influence Pruning in Multilingual Language Models
arXiv 2610.21383safety and guardrails Published 2026-10-15
Evaluating and Improving Static Application Security Testing Tools for Rust
arXiv 2610.21373coding agents Published 2026-10-15
Agentic Dependency Transformations: Lessons from an Exploratory Case Study of Flask
arXiv 2610.21372agent evaluation Published 2026-10-15
SWE-at-Scale: Scalable Issue Synthesis from Historical Bug Fixes
arXiv 2610.21359sparsity and pruning Published 2026-10-15
PSA: A Pruning-Friendly Sparse-Aware Accelerator for Efficient Sparse Convolution in SNNs
arXiv 2610.21313safety and guardrails Published 2026-10-15
Combining LLM and Dynamic Analysis for Universal Testing of Static Application Security Testing Tools
arXiv 2610.21309agent evaluation Published 2026-10-15
VALCON: Benchmarking and Enhancing Value Consistency in Large Language Models via Dual-Role Variation
arXiv 2610.21304coding agents Published 2026-10-15
Mallea: Building, Steering, and Memorizing Worlds with Coding Agent
arXiv 2610.21283tool use Published 2026-10-15
TRACER: Trajectory-Aware Tool Retrieval for LLM Agents
arXiv 2610.21267tool use Published 2026-10-15
TRACER: A Multimodal Harness for Patient Care in Hospitals
arXiv 2610.21266attention and kernels Published 2026-10-15
AdaBool: Sparse Boolean Matrix Processing with Adaptive Data Representation
arXiv 2610.21264quantization Published 2026-10-15
Fractional Power-of-Two Quantization for Efficient and Effective Multiplier-Free LLM Inference
arXiv 2610.21257multi-agent orchestration Published 2026-10-15
AutoOnboarder: Autonomously Onboarding LLM Agents
arXiv 2610.21218efficient training Published 2026-10-15
Hard Way, The Only Way: Provable Merits of Training on Hardest Examples
arXiv 2610.21201agent evaluation Published 2026-10-15
AgentEval: Evidence-Bounded Assessment of Stateful Agents
arXiv 2610.21177memory and context Published 2026-10-15
MemScope: Evolving Agent Memory Architectures through Localized Code Repair
arXiv 2610.21170efficient training Published 2026-10-15
GRADS: Graph-Regulated Agentic Data Synthesis with Controllable Difficulty
arXiv 2610.21167memory and context Published 2026-10-15
From Long Context to Selective Memory:A Survey of Memory-Centric Compression for Large Language Models
arXiv 2610.21165agent evaluation Published 2026-10-15
Stronger Judges Do Not See More: Deployment-Integrity Blindness in LLM Evaluation
arXiv 2610.21135KV cache Published 2026-10-15
Compression Does Not End at Selection: Post-Selection Positional Geometry for Sparse KV Caches
arXiv 2610.21126multi-agent orchestration Published 2026-10-15
SSR: Structural Signal Retriever for Scalable Agent Discovery in LLM-Orchestrated Multi-Agent Systems
arXiv 2610.21113agent evaluation Published 2026-10-15
Kolab: Benchmarking Human-Agent Collaboration in Long-Horizon Workflows
arXiv 2610.21109agent evaluation Published 2026-10-15
ToolBottleneckBench: Measuring Tool-Agent Performance under Controlled Pressures and Fixed Harness Budgets
arXiv 2610.21098other Published 2026-10-15
Helix: A Unified Programmable Accelerator for Diverse Krylov Subspace Solvers
arXiv 2610.21096memory and context Published 2026-10-15
MemComposer: Learning to Assemble Memories for LLM Agents via Reinforcement Learning
arXiv 2610.21091agent evaluation Published 2026-10-15
Probe-and-Resume: When and How to Evaluate Agent Execution
arXiv 2610.21088agent evaluation Published 2026-10-15
Sci$Π$vot: Benchmarking Principal-Investigator Judgment at Pivotal Moments of Scientific AutoResearch Agents
arXiv 2610.21082coding agents Published 2026-10-15
Skip the Compressor: Coding Agents Fix More Issues When They Pick Whole Files
arXiv 2610.21073other Published 2026-10-15
Probabilistic Optical-Layer Surrogate for Risk-Aware Control-Plane Decision Making
arXiv 2610.21069other Published 2026-10-15
SALTO: Solver-Anchored Lifted Transfer Operators for Nonlinear Event Inference
arXiv 2610.21049agent evaluation Published 2026-10-15
Learning When to Withdraw: Curriculum Coverage and Rule-Specific Failures in Fine-Tuned Language Agents
arXiv 2610.21045agent evaluation Published 2026-10-15
SCARCE: Emergent Institutions in Long-Horizon Multi-Agent Coding Systems
arXiv 2610.21029safety and guardrails Published 2026-10-14
StepSafe: Learning Timely Intervention with Context-Aware Safety Reasoning
arXiv 2610.20960memory and context Published 2026-10-14
Same Context, Different Agents: A Behavioral Study of Semantic Layers, Orchestration and Compaction in Data Agents
arXiv 2610.20959agent evaluation Published 2026-10-14
MAI-Gym: An Open All-in-One Platform for GUI Agent Evaluation and RL Training across Mobile and Computer Environments
arXiv 2610.20939memory and context Published 2026-10-14
Prune the Summary, Not the History: Source-Judged Compaction for Long-Horizon Agents
arXiv 2610.20927memory and context Published 2026-10-14
SkillNEO: Non-Elitist Evolutionary Optimization of Agent Skills
arXiv 2610.20926other Published 2026-10-14
Flexible Doubly Robust Proximal Causal Inference for Censored Outcomes
arXiv 2610.20893speculative decoding Published 2026-10-14
Long-Context Drafting Does Not Require Long-Context Drafters
arXiv 2610.20865quantization Published 2026-10-14
Task-Phase-Conditioned Expert Routing Is Not a Reliable Proxy for Quantization Sensitivity in MoE FFN Tensors
arXiv 2610.20826safety and guardrails Published 2026-10-14
PRISM: Transfer and Limits of Activation Interventions in Tool-Calling Language Models
arXiv 2610.20808other Published 2026-10-14
Sliced Wasserstein Sketches with Error Certificates
arXiv 2610.20788failure recovery Published 2026-10-14
ARIA: Runtime Inspection and Amendment for Multi-Agent Systems
arXiv 2610.20780safety and guardrails Published 2026-10-14
INFUSE: A Unified Surface Benchmark for Indirect Prompt Injection in Agentic Systems
arXiv 2610.20776speculative decoding Published 2026-10-14
On-Policy Residual Learning for Diffusion Language Models
arXiv 2610.20687KV cache Published 2026-10-14
RoLaQ: Joint Rotation and Range Calibration for 2-Bit KV Caches in Multi-Head Latent Attention
arXiv 2610.20686agent evaluation Published 2026-10-14
HPCAgent-Bench: A Framework for Sound Benchmarking of Code-Optimizing Agents, Harnesses, and Skills
arXiv 2610.20673agent evaluation Published 2026-10-14
Evaluating Model Capabilities After Benchmark Saturation
arXiv 2610.20667sparsity and pruning Published 2026-10-14
Your Transformer Is Secretly Elastic
arXiv 2610.20588efficient training Published 2026-10-14
Player-1: Learning to Play from Synthetic Experience
arXiv 2610.20584KV cache Published 2026-10-14
Latent Compaction with Test-Time Training
arXiv 2610.20582tool use Published 2026-10-14
One Task, Many Interfaces: Simple Cross-Interface Consistency Training Improves LLM Agent Robustness
arXiv 2610.20576multi-agent orchestration Published 2026-10-14
Scaling Discovery Requires Forgetting
arXiv 2610.20563other Published 2026-10-14
A Two-Level Clinical Novelty Prioritization Framework for Disease Trajectories Discovered from Longitudinal Electronic Health Records
arXiv 2610.20512speculative decoding Published 2026-10-14
ReaDy: Recurrent States as Dynamic Context for Speculative Drafting in Hybrid Linear Attention Large Language Models
arXiv 2610.20503efficient training Published 2026-10-14
AdaRSI: On-Demand Recursive Self-Improvement for Embodied Agents
arXiv 2610.20474other Published 2026-10-14
Selective Growing Q-Networks (SGQN): An Analysis of Adaptive Action Grids in Decoupled Q-Learning
arXiv 2610.20463memory and context Published 2026-10-14
The COUNT: Counting N-grams Unlocks Scaling on Host Memory and Training-Free Adaptation
arXiv 2610.20456safety and guardrails Published 2026-10-14
From Training to Escalation: Forensic Analysis and Local Recreation of RL/RLHF-Induced Privilege Escalation in AI Agents (CVE-2026-65616)
arXiv 2610.20437other Published 2026-10-14
HRTF Upsampling and Personalization via Neural Operators
arXiv 2610.20400memory and context Published 2026-10-14
Cortex: Consolidating Context into Parametric Memory
arXiv 2610.20399speculative decoding Published 2026-10-14
FOLD: Position-Free Far Memory with Local Order for Long-Context Speculative Decoding
arXiv 2610.20392tool use Published 2026-10-14
From Relevance to Validity: Zero-Shot Tool Selection via Functional and Contextual Validation
arXiv 2610.20388other Published 2026-10-14
Understanding NVIDIA Grace-Blackwell Unified Memory Through Placement-Controlled Microbenchmarks
arXiv 2610.20354quantization Published 2026-10-14
Beyond Expert Rankings: Contrast Calibration for Mixed-Precision MoE Quantization
arXiv 2610.20353other Published 2026-10-14
Comparative Evaluation of Deep-Learning based DoA Estimators for Real-Time ISAC Beamforming
arXiv 2610.20326quantization Published 2026-10-14
SpectrumQuant: Spectrum Alignment for LLM Quantization
arXiv 2610.20303coding agents Published 2026-10-14
Long-Horizon RL Needs Muon
arXiv 2610.20256other Published 2026-10-14
Repairing Speech Access After Text Updates with Frozen Affine Fields
arXiv 2610.20253coding agents Published 2026-10-14
P2Patch: Exploit-Guided Vulnerability Repair
arXiv 2610.20245efficient training Published 2026-10-14
Continual pretraining Prostate MRI ViT Encoder for task alignment
arXiv 2610.20217memory and context Published 2026-10-14
HEM-Med: A Hierarchical Agentic Memory Framework for Safe Multi-Center Medication Recommendation
arXiv 2610.20191speculative decoding Published 2026-10-14
TreeSpark: Learning to Draft the Trees You Verify for Speculative Decoding
arXiv 2610.20175distillation Published 2026-10-14
RM-OPD: On-Policy Distillation for Generative Reward Models
arXiv 2610.20174other Published 2026-10-14
Optimizing Sparse Boolean Rule Networks using Mixed Integer Linear Programming
arXiv 2610.20170other Published 2026-10-14
Sparse Autoencoders for Mechanistic Interpretability in NLP: A Critical Survey of Recoverability, Interpretability, and Utility
arXiv 2610.20139serving and batching Published 2026-10-14
Seeing at the Edge: Energy Characterization of Vision-Language Model Inference on Edge Platforms
arXiv 2610.20136other Published 2026-10-14
KANLab: TensorFlow Playground for Kolmogorov-Arnold Networks
arXiv 2610.20117memory and context Published 2026-10-14
Maximising Transparency in Context Compaction Improves Retention of Facts and Commitments for Lightweight Coding Agents
arXiv 2610.20116efficient architectures Published 2026-10-14
Why Generate a Click? Plug-in Decision Heads for Web Agents
arXiv 2610.20107agent evaluation Published 2026-10-14
EvalBuilder: Can LLMs Build Efficient Benchmarks?
arXiv 2610.20099other Published 2026-10-14
Designing Textual Interfaces To Enhance Llm Reasoning On Audio-Visual Question Answering
arXiv 2610.20085coding agents Published 2026-10-14
The Code of Theseus: Can Agents Rebuild Software from Recovered Design Documents?
arXiv 2610.20068agent evaluation Published 2026-10-14
What Coding Agents Leave Unsaid: Measuring Task-State Gaps Between Execution and Reports
arXiv 2610.20003multi-agent orchestration Published 2026-10-14
A Strong Harness Narrows the Model Gap: A Comparative Evaluation for Legislative Document-to-XML Conversion
arXiv 2610.19999other Published 2026-10-14
STISA: A 0.16 GOPS/W/PE Single-Shot Inference FPGA-based SNN Accelerator with Algorithm and Hardware Co-Design
arXiv 2610.19973efficient architectures Published 2026-10-14
Choosing the Number of Attention Heads from Intrinsic Dimension: A Small-Scale Study
arXiv 2610.19966sparsity and pruning Published 2026-10-14
To Reason but Not to Stop: Understanding and Recovering Pruned Large Reasoning Models
arXiv 2610.19934efficient training Published 2026-10-14
Explorer-Zero: Improving Self-Play for Long-Horizon Tasks with Decomposable Verifiable Rewards
arXiv 2610.19868multi-agent orchestration Published 2026-10-14
When Extremists Enter LLM Societies: Investigating Extreme Behavior in Multi-Agent Systems
arXiv 2610.19863efficient architectures Published 2026-10-14
Energy Consumption Analysis of Discrete Diffusion and Autoregressive Language Models
arXiv 2610.19855multi-agent orchestration Published 2026-10-14
LAGO: Language Agents with Reusable Growing Options
arXiv 2610.19843other Published 2026-10-14
CTIA-ODE: Continuous-Time Inter-Variable Attention for Long-Term Time Series Forecasting
arXiv 2610.19831distillation Published 2026-10-14
ActionOPD: Distilling Decisions, Not Tokens
arXiv 2610.19820serving and batching Published 2026-10-14
Delta Tokenizer: Reference-Identical Tokenization of Session Deltas and Segment-Parallelism for Long Prompts on Multicore CPUs
arXiv 2610.19801speculative decoding Published 2026-10-14
JetSpec++: Foresight Parallel Drafting with Self Refinements
arXiv 2610.19792other Published 2026-10-14
Computational compressive imaging reconstruction with a simulated-annealing-based Ising solver
arXiv 2610.19778safety and guardrails Published 2026-10-14
Analyzing and Mitigating Poisoning Attacks on Flow-Based Intrusion Detection
arXiv 2610.19766quantization Published 2026-10-14
Discriminative Score Function: Turning Pretrained Models into Functional Generative Priors
arXiv 2610.19743efficient training Published 2026-10-14
ReGRA: Residual-Gradient Adaptation for Complementary Low-Rank Fine-Tuning
arXiv 2610.19700other Published 2026-10-14
CAFE: Conducting-Cell-Aware Adaptive Differential Encoding for Robust DNN Inference in FeFET-Based Compute-in-Memory
arXiv 2610.19697sparsity and pruning Published 2026-10-14
ESL-former: Event Sparse Linear Transformer for FPGA-Accelerated Malware Detection
arXiv 2610.19665efficient architectures Published 2026-10-14
DAGFormer: Global Topology Prediction for Transformer Cross-Layer Connections
arXiv 2610.19625KV cache Published 2026-10-14
PN-Mix: Pair-Norm Guided Mixed-Precision Key Cache Quantization for Efficient Long-Output LLM Inference
arXiv 2610.19605safety and guardrails Published 2026-10-14
What You Sign Is Not What You Run: The Verifiability Gap in Quantized LLM Supply Chains
arXiv 2610.19498distillation Published 2026-10-14
Progress-Aware Distillation for Mitigating Stagnation in Small Language Model Agents
arXiv 2610.19439memory and context Published 2026-10-13
Titan-Llama: Retrofitting Pre-trained LLMs with External Neural Memory Modules
arXiv 2610.19210sparsity and pruning Published 2026-10-13
Adaptive Computation with Conformal Certification
arXiv 2610.19189sparsity and pruning Published 2026-10-13
Unlocking Structured Pruning for Reasoning Language Models via Attention-Preserving MLP Trimming
arXiv 2610.19121efficient training Published 2026-10-13
How Many Ranks Does a Token Need? A Noise-Floor Squelch for Rank-One LoRA Experts
arXiv 2610.19033KV cache Published 2026-10-13
Joint Subspace Sparse Attention for Efficient Long-Context Decoding
arXiv 2610.19026other Published 2026-10-13
Discrete Diffusion Models are Minimax Optimal Under Sparsity
arXiv 2610.19020safety and guardrails Published 2026-10-13
Discovering Undisclosed Concepts in LoRA Adapters for Diffusion-Based AIGC Models
arXiv 2610.18990other Published 2026-10-13
Delay Complexity Dimension: Learning with Delayed and Censored Labels
arXiv 2610.18935planning and decomposition Published 2026-10-13
ChunkAct: Action Chunking for Agentic Robotics
arXiv 2610.18874speculative decoding Published 2026-10-13
DoubtLess: Training-Free Contrastive Decoding for Efficient Reasoning
arXiv 2610.18862tool use Published 2026-10-13
Hybrid Reasoning Agents
arXiv 2610.18861coding agents Published 2026-10-13
The Last Remote Sensing Model Built by Humans? Casting Any Earth Observation Task as Recursive Self-Improvement
arXiv 2610.18845other Published 2026-10-13
RoboSimStudio: An Interactive Toolkit for Rapid Real2Sim of Robot Manipulation Tasks
arXiv 2610.18835agent evaluation Published 2026-10-13
SEAR: Self-Evolving Agentic Robotics
arXiv 2610.18792agent evaluation Published 2026-10-13
SolBehave: Evaluating the Behavioural Correctness of LLM-Generated Smart Contracts
arXiv 2610.18745speculative decoding Published 2026-10-13
SP$^2$ec: Adaptive Self-Speculative Decoding for Vision-Language Models
arXiv 2610.18735memory and context Published 2026-10-13
Just Respect the Semantics: Image-Token Compression via Pre-LLM Semantic Variable-Resolution Tokenization and In-LLM Relevance Pruning
arXiv 2610.18690safety and guardrails Published 2026-10-13
Knowing the User Is Not Knowing What We Established: Grounding Awareness in Personal Agents
arXiv 2610.18684efficient training Published 2026-10-13
ProAS: A Progress-Aware Advantage Shaping Framework for Multi-Turn Agentic Reinforcement Learning
arXiv 2610.18683other Published 2026-10-13
Model-Based Online Decision Making via Generative Trajectory Planning
arXiv 2610.18677KV cache Published 2026-10-13
Kwai Summary Attention V2 Technical Report
arXiv 2610.18654KV cache Published 2026-10-13
MultiAlloc: Budget-Aware Precision Allocation for KV Cache Quantization
arXiv 2610.18642tool use Published 2026-10-13
Knowledge Graphs as Files: An Agent Harness Where Tool Creation and Tool Use Co-Evolve
arXiv 2610.18616agent evaluation Published 2026-10-13
Can an Agent Learn to Predict Its Own Failure? Outcome-Grounded Confidence Calibration in LLM Agents
arXiv 2610.18614distillation Published 2026-10-13
Does self-distillation improve reasoning?
arXiv 2610.18613agent evaluation Published 2026-10-13
Reading Right, Acting Wrong: Diagnosing Decision Failures of System One Models in Agent Loops
arXiv 2610.18579tool use Published 2026-10-13
Lost in Implicit Routing: How LLM Agents Fail to Load Relevant Skills at Scale
arXiv 2610.18522sparsity and pruning Published 2026-10-13
Two For The Price Of One: Curvature-Aware Pruning Preserves Influence Functions Too
arXiv 2610.18516coding agents Published 2026-10-13
The Last Mile of Coding Agents for Long-Document Tasks: An Empirical Study on Clinical Trial Protocol Digitization
arXiv 2610.18507agent evaluation Published 2026-10-13
Judges Follow the Reference: The Derivation Gap in Verifying Tool-Call Arguments
arXiv 2610.18497other Published 2026-10-13
ProTok: Unified Protein Tokenization via Modular Multi-Codebook Quantization
arXiv 2610.18481sparsity and pruning Published 2026-10-13
Readout Scaling Shapes Sparse Feature Recovery
arXiv 2610.18462tool use Published 2026-10-13
Teacher-Guided Tool Learning for Small Language Models via Bidirectional KL Constraints
arXiv 2610.18459tool use Published 2026-10-13
Urbex: Agentic Spatial Grounding for City-Scale 3D Scenes
arXiv 2610.18356planning and decomposition Published 2026-10-13
Let the Mind Conceive, Let the Hand Achieve: Following Multimodal Instructions for Long-Horizon Robotic Manipulation
arXiv 2610.18352safety and guardrails Published 2026-10-13
Jev-Guard: Fast Routing and Auditing Agent Actions over Editable Policy Trees
arXiv 2610.18326agent evaluation Published 2026-10-13
SimInter: Evidence-Grounded User Simulation with Interview-Based Profile Enrichment for Conversational Recommender Systems
arXiv 2610.18308KV cache Published 2026-10-13
SparPR: Runtime KV-Cache Recovery for Resource-Constrained LLM Inference on Consumer Hardware
arXiv 2610.18140quantization Published 2026-10-13
Revisiting Error Feedback for Model Quantization
arXiv 2610.18139other Published 2026-10-13
H-DDCM: Hadamard Codebooks for Fast Generative Image and Video Compression
arXiv 2610.18125KV cache Published 2026-10-13
ChampionKV: Head-Wise Adaptive Budget Allocation for KV Cache Compression
arXiv 2610.18026KV cache Published 2026-10-13
LatentRemedy: Efficient Remedy For Non-Prefix KV Cache Reuse in Multi-Agent Systems
arXiv 2610.18014other Published 2026-10-13
Exact power indices for plurality-voting ensembles
arXiv 2610.17980KV cache Published 2026-10-13
AlignPred: Error-Controlled KV Cache Compression via Alignment and Predictive Coding
arXiv 2610.17970efficient training Published 2026-10-13
Neural Dimensional Analysis: Foundations of Dimensionless and Scale-Consistent Computation
arXiv 2610.17969KV cache Published 2026-10-13
BELT : BEhavior-Aligned Low-Rank KV Cache Compression with Learned Top-k Rank Allocation
arXiv 2610.17966KV cache Published 2026-10-13
Virtual Head Attention
arXiv 2610.17959serving and batching Published 2026-10-13
Batch-Paced: Fixed-Budget Batch Decoding in Discrete Diffusion Language Models
arXiv 2610.17917safety and guardrails Published 2026-10-09
X-AgentRed: Red-Teaming Personal Agents in Evolving Multi-Service Ecosystems
arXiv 2610.13886failure recovery Published 2026-10-09
Returning to Development States: When Full Sandbox Snapshots Help Coding Agents
arXiv 2610.13841sparsity and pruning Published 2026-10-09
FORTE: A Production Framework for Feature Evaluation and Deprecation in Industrial Recommender Systems
arXiv 2610.13806agent evaluation Published 2026-10-09
OccludeBench: Can VLMs Find What's Missing?
arXiv 2610.13790serving and batching Published 2026-10-09
ReplaySSM: Cache SSM Inputs, Not State
arXiv 2610.13778quantization Published 2026-10-09
DualQuant: Dual-scale Calibration-free Weight-and-Activation Quantization for LLMs
arXiv 2610.13767KV cache Published 2026-10-09
Adaptive DHiSparse: Speculative Decoding over a Load-Adaptive KV Hierarchy
arXiv 2610.13708failure recovery Published 2026-10-09
Inject a Turn, Not a Token: Learning Where to Redirect LLM Reasoning for Self-Correction
arXiv 2610.13672coding agents Published 2026-10-09
(When) Should a Software Factory be Lights-off? Benchmarking Code Review Agents on Bug Detection
arXiv 2610.13668multi-agent orchestration Published 2026-10-09
Humanize Olympiad Agents: A Multi-Session Workflow for Olympiad Problem Solving with Solve, Review, Wiki, Verify
arXiv 2610.13632multi-agent orchestration Published 2026-10-09
Humanize MLE Agents: Multi-Agent Session Alternation Beyond Single-Agent Limits
arXiv 2610.13494multi-agent orchestration Published 2026-10-09
Proactive Agentic Orchestration
arXiv 2610.13485attention and kernels Published 2026-10-09
CompileMask: Bridging the Gap Between Expressiveness and Efficiency in Attention Masking
arXiv 2610.13474other Published 2026-10-09
Towards Enriching Design-Time Models with Evidence from Simulation and Operational Activities
arXiv 2610.13456efficient training Published 2026-10-09
Rollout-Time Intervention: Unlocking Agentic Reinforcement Learning Beyond Policy Limits
arXiv 2610.13438tool use Published 2026-10-09
Web$hell: A Unified Interface for Agentic Web Access
arXiv 2610.13416other Published 2026-10-09
ShoreKeeper: Accelerating Iterative Data Analytics in the Cloud via Distributed Recovery-Cost-Aware Caching
arXiv 2610.13396coding agents Published 2026-10-09
ProgramGym: Scaling Training Environments for Long-Horizon Program Reproduction Agents
arXiv 2610.13395agent evaluation Published 2026-10-09
AgentSWE: Can Coding Agents Build the Agent You Actually Want?
arXiv 2610.13361agent evaluation Published 2026-10-09
SWE-Critix: Post-Training Generalist Verifiers for Assessing Coding-Agent Trajectories
arXiv 2610.13356tool use Published 2026-10-09
Masked Diffusion Language Agents for Tool-Integrated Chemical Reasoning
arXiv 2610.13330retrieval and RAG Published 2026-10-09
PRIME: Pre-Retrieval Index Modeling Enhancement for Structured Document Retrieval-Real
arXiv 2610.13296retrieval and RAG Published 2026-10-09
Beyond the Final Embedding: Retrieving Memories from Token-Level Evidence
arXiv 2610.13277multi-agent orchestration Published 2026-10-09
Agent4Spec: An LLM-based Agentic System to Optimize API Specifications through Automated Validation and Market Analysis
arXiv 2610.13273safety and guardrails Published 2026-10-09
Forgetting is Not Always Bad: A Neuro-Inspired Memory Repair Mechanism for Poisoned LLM Agents
arXiv 2610.13267safety and guardrails Published 2026-10-09
Understanding and Mitigating Cross-Session Attacks via Persistent Artifacts in Agent Skill Composition
arXiv 2610.13227efficient training Published 2026-10-09
Monarch-MoE: Structured Experts for Multimodal Continual Instruction Tuning
arXiv 2610.13215sparsity and pruning Published 2026-10-09
How Much Computation Does Every Forecast Need? Structured Block Skipping in Two Time-Series Foundation Models
arXiv 2610.13208speculative decoding Published 2026-10-09
S$^3$R: Fast and Faithful RL Rollouts via Self-Sparse Speculative Decoding
arXiv 2610.13202planning and decomposition Published 2026-10-09
JANUS: Enhancing Long-Horizon Autonomous Research with Co-Evolving World Models
arXiv 2610.13194planning and decomposition Published 2026-10-09
RoleWorld: Knowledge Transfer in World Modeling with Functional Role Abstractions
arXiv 2610.13172coding agents Published 2026-10-09
Scaling Agentic Data for Long-Horizon Terminal Intelligence
arXiv 2610.13166coding agents Published 2026-10-09
Quality, Not Just Outcome: A Structural Scoring Framework for Selecting Coding-Agent Trajectories for Supervised Fine-Tuning
arXiv 2610.13152safety and guardrails Published 2026-10-09
DRYLAB-Bench: Benchmarking the Biosecurity Risk of Agentic LLM Mutation Design
arXiv 2610.13116memory and context Published 2026-10-09
LongHarness: Advancing Agentic Reasoning over 100M-Token Corpora via Stateful Evidence Management
arXiv 2610.13075other Published 2026-10-09
TPCS: A Cycle-Level Simulation Paradigm with Native Execution Semantics for RTL Modeling and Generation
arXiv 2610.13067agent evaluation Published 2026-10-09
TestSuiteBench: Coding Agents Struggle to Generate Strong Test Suites from Specifications
arXiv 2610.13065KV cache Published 2026-10-09
KV-Repurpose: A Testbed for Evaluating Non-Prefix KV-Cache Reuse
arXiv 2610.13060quantization Published 2026-10-09
RAMP: Recurrent Adaptive Mixed Precision for Efficient Prefill in Recurrent-Depth Language Models
arXiv 2610.13041efficient architectures Published 2026-10-09
Lightweight Time-Frequency Vibration Sensing for Consumer Device Condition Monitoring
arXiv 2610.13040other Published 2026-10-09
Low-Rank Kolmogorov-Arnold Generator Flows for Interpretable Operator Learning
arXiv 2610.13028coding agents Published 2026-10-09
Evo$^2$: Evolving the Evolver through Weight-Harness Co-Evolution
arXiv 2610.13025agent evaluation Published 2026-10-09
Don't Guess, Just Ask: Evaluating Interactive Requirement Clarification for Repository-Level Software Development
arXiv 2610.13022agent evaluation Published 2026-10-09
EmbodiedCLBench: Evaluating Continual Learning for Self-Evolving Embodied Agents
arXiv 2610.13021planning and decomposition Published 2026-10-09
Work2Skill: Capturing Human GUI Experience for Reusable Agent Skill Induction
arXiv 2610.12984speculative decoding Published 2026-10-09
Surfing the Activation Wave: Analyzing the Information and Geometry of LLM Hidden States for Speculative Decoding
arXiv 2610.12939safety and guardrails Published 2026-10-09
Vulnerable Models in the AI Supply Chain: An Empirical Study of Adoption, Propagation, and Usage Patterns
arXiv 2610.12893other Published 2026-10-09
Scalable Learning of Determinantal Point Processes with Kernel-Structured Interactions
arXiv 2610.12829safety and guardrails Published 2026-10-08
VEPA: Verified Evolving Formal Policy for Lifelong LLM Agent Guardrails
arXiv 2610.12813other Published 2026-10-08
The Expressive Power of Random-Feature MPNNs Across Graph Sizes
arXiv 2610.12809memory and context Published 2026-10-08
EvoBelief: Falsifiable World Models for Self-Improving LLM Agents
arXiv 2610.12789agent evaluation Published 2026-10-08
MADS: Multi-Agent Dialog Simulation for Robust Conversational Agents
arXiv 2610.12757agent evaluation Published 2026-10-08
How Do Hardware Faults Manifest Differently in LLM Agents?
arXiv 2610.12756sparsity and pruning Published 2026-10-08
Zebra3D: Efficient Hybrid Foundation Models For Scalable 3D Reconstruction
arXiv 2610.12727safety and guardrails Published 2026-10-08
Byzantine-Resilient Federated Large Language Model via Hypergraph Message Passing
arXiv 2610.12711memory and context Published 2026-10-08
Recurrent Harness for Agentic Continual Learning
arXiv 2610.12668KV cache Published 2026-10-08
Kronecker Product Attention: Memory-Efficient KV Compression via Structured Factorization
arXiv 2610.12657other Published 2026-10-08
CLEAT: Learning Value-Guided Clarification for Interactive Agents
arXiv 2610.12655agent evaluation Published 2026-10-08
SoftwareGym: Scaling Computer-Use Agent Training with Diverse Synthesized Software Environments
arXiv 2610.12611agent evaluation Published 2026-10-08
SpireBench: Evaluating Long-Horizon Strategic Play of Agentic Models with Slay the Spire
arXiv 2610.12596efficient architectures Published 2026-10-08
One Model to Loop Them All: Elastic Latent Recurrence for Language Models
arXiv 2610.12582speculative decoding Published 2026-10-08
Contextual Correction of Parallel Draftsfor Speculative Decoding
arXiv 2610.12570agent evaluation Published 2026-10-08
SWE-Duel: Towards a Self-Scaling Arena for Coding Agents
arXiv 2610.12569other Published 2026-10-08
Behind-the-Meter Photovoltaic Visibility from Net Smart Meters: Deployment Regimes and Operational Insights
arXiv 2610.12527agent evaluation Published 2026-10-08
BrickBench: Evaluating Agentic Brick Design
arXiv 2610.12452safety and guardrails Published 2026-10-08
Caught in the Act: Probes Effectively Detect Sabotage and Catch Unverbalized Deception
arXiv 2610.12445efficient training Published 2026-10-08
Rounding in Preconditioner Space: Redesigning 4-bit AdamW Optimizer-State Quantization
arXiv 2610.12444failure recovery Published 2026-10-08
OnTrack: Real-Time Monitoring and Intervention in LLM Agent Trajectories via Streaming Structure-Aware Optimal Transport
arXiv 2610.12375agent evaluation Published 2026-10-08
Accurate but Not Humble: Evaluating Epistemic Humility in LLM Agents under Knowledge Conflict
arXiv 2610.12360KV cache Published 2026-10-08
VFold: Symmetry-Aware Cross-Layer Value Cache Compression
arXiv 2610.12338other Published 2026-10-08
asdex: Automatic Sparse Differentiation in JAX
arXiv 2610.12336sparsity and pruning Published 2026-10-08
SparseDecoding: Decoding-Aware Pruning for Accurate and Efficient LLM Inference
arXiv 2610.12327safety and guardrails Published 2026-10-08
One Word Opens the Gate: The Option-Channel Attack on Typed Decision Models as Agent Guardrails
arXiv 2610.12292agent evaluation Published 2026-10-08
TestPrism: Rethinking Test Evaluation Beyond a Single Reference
arXiv 2610.12289tool use Published 2026-10-08
Unlocking the Regulatory Genome by ARGUS: An Evidence-Constrained Agentic Framework for Interpreting Single Nucleotide Variants
arXiv 2610.12281coding agents Published 2026-10-08
Cadence: Strategic Guidance for Coding Agents
arXiv 2610.12269efficient training Published 2026-10-08
AdaCast: Conditional Parameter Generation for Adaptive Time Series Forecasting
arXiv 2610.12240agent evaluation Published 2026-10-08
A Closer Look at Agentic BBO: Benchmarking LLM Agents for Black-Box Optimization
arXiv 2610.12183multi-agent orchestration Published 2026-10-08
Recursive Self-Improvement through Multi-Agent Self-Supervision
arXiv 2610.12176KV cache Published 2026-10-08
Rehearse Everything, Remember Nothing: Attic-KV Rehearses What Will Be Read
arXiv 2610.12133coding agents Published 2026-10-08
EvoAlloc: A Self-Evolving Resource Allocation Agent for Efficient Program Evolution
arXiv 2610.12086planning and decomposition Published 2026-10-08
When Should Agents Think? Adaptive Reasoning via Cross-Turn Estimation
arXiv 2610.12061quantization Published 2026-10-08
The Polytopal Neural Network
arXiv 2610.12004agent evaluation Published 2026-10-08
Can LLMs Fix It Without Code? Toward Automated Verification of No-Code Bug Fixes
arXiv 2610.11963efficient training Published 2026-10-08
MiMo-V2.6: Scaling Reinforcement Learning Towards Self-Improvement
arXiv 2610.11959memory and context Published 2026-10-08
Event-Centric Memory with Query-Aware Graph Augmentation for Long-Term Conversational Agents
arXiv 2610.11920other Published 2026-10-08
RobustLDS: Learning linear dynamical systems under adversarial corruptions
arXiv 2610.11906multi-agent orchestration Published 2026-10-08
Forms of LLM-Integrated Applications from LLM-Chats to Autonomous AI Agent System
arXiv 2610.11899coding agents Published 2026-10-08
Trajectory-Guided Fault Localization for Agent Skill Evolution
arXiv 2610.11858sparsity and pruning Published 2026-10-08
DADP: Dynamic Activity-Dependent Pruning, A Reverse Hebbian-Inspired Structural Pruning Method
arXiv 2610.11853memory and context Published 2026-10-08
Memento 3: Model-Based Recursive Self-Improvement through Reflective Rulebooks
arXiv 2610.11794efficient architectures Published 2026-10-08
Easy to anticipate, hard to compute: boundary dependence finds the computed outputs that entropy patching misses
arXiv 2610.11790KV cache Published 2026-10-08
Compile the Table: Query-Calibrated Operator Compression for Tabular In-Context Learning
arXiv 2610.11784memory and context Published 2026-10-08
Narrow and Deep: An Ontology Tower as the Knowledge of an LLM Agent for an Industrial Equipment System
arXiv 2610.11768agent evaluation Published 2026-10-08
What Output-Only Review Cannot Verify: Study Contracts for Research Agents
arXiv 2610.11754other Published 2026-10-08
DEX: Digit-Level Early Exit for Energy-Efficient MSDF Neural Network Inference
arXiv 2610.11748memory and context Published 2026-10-08
MemTrial: Learning When to Trust Memory in LLM Portfolio Agents
arXiv 2610.11732agent evaluation Published 2026-10-08
TRACE: Diagnosing Verifier Brittleness in Agentic Evaluation
arXiv 2610.11678efficient training Published 2026-10-08
Harness Evolution Hits a Ceiling: When Weight Training Should Begin
arXiv 2610.11655coding agents Published 2026-10-08
One Skill Too Many: How Co-Installed Skills Conflict in Coding Agents
arXiv 2610.11647agent evaluation Published 2026-10-08
PolyCodeEval: Benchmarking Multilingual Code Generation from Functions to Repositories
arXiv 2610.11618memory and context Published 2026-10-08
Where Do the Tokens Go? Understanding and Reducing Costs in LLM Agents for Vulnerability Discovery
arXiv 2610.11602agent evaluation Published 2026-10-08
Error-Propagation Modeling for Failure Attribution in LLM-Based Multi-Agent Systems
arXiv 2610.11600coding agents Published 2026-10-08
Runnable Commit Untangling for Coding Agents
arXiv 2610.11593memory and context Published 2026-10-08
Chronos Enables Code Agents to Reason over Software Evolution
arXiv 2610.11578agent evaluation Published 2026-10-08
SWE-Journey: Towards More Realistic Evaluation of Coding Assistants through Long-Horizon, Multi-Turn Interaction
arXiv 2610.11559multi-agent orchestration Published 2026-10-08
Safe, Persistent, and Evolving Agent Harness for Understanding Partially Observable Worlds
arXiv 2610.11552distillation Published 2026-10-08
ReTeach: Building a Self-Teacher through Multi-Round Reflection and Retry
arXiv 2610.11529agent evaluation Published 2026-10-08
SSCBench: Evaluating the Evidential Validity of Fault-Injection Tests for Tool-Using LLM Agents
arXiv 2610.11514other Published 2026-10-08
Compactness and Consistency: A Conjoint Framework for Deep Graph Clustering
arXiv 2610.11506agent evaluation Published 2026-10-08
Evaluating Local Language Model Agents for Reproducible Data Engineering: An Empirical Software Engineering Study of Mobility Workflows
arXiv 2610.11482multi-agent orchestration Published 2026-10-08
RoboAware: Learning to Coordinate Embodied Skills from Counterfactual Outcomes
arXiv 2610.11480multi-agent orchestration Published 2026-10-08
Generative Adversarial Loops
arXiv 2610.11458agent evaluation Published 2026-10-08
Tracing the Thoughts of a Coding Agent Playing ARC-AGI-3: Lessons for Continual Learning
arXiv 2610.11450agent evaluation Published 2026-10-08
Closed-loop evaluation of LLM agents for embedded software development
arXiv 2610.11447retrieval and RAG Published 2026-10-08
RIT-RAG: Navigating Document Corpora with Retrieval-Induced Trees
arXiv 2610.11370KV cache Published 2026-10-08
RaReCache: Bridging the Gap in Cross-Model KV Cache Reuse via Rank disagreement-based Selective Recomputation
arXiv 2610.11358safety and guardrails Published 2026-10-08
Deception by Omission: Language Models Knowingly Hide Their Mistakes
arXiv 2610.11351efficient training Published 2026-10-08
SynCo: Data Synthesis Co-Training for Self-Evolving LLMs via Multi-Agent Reinforcement Learning
arXiv 2610.11345multi-agent orchestration Published 2026-10-08
EvoSim: Learning to Model, Modeling to Learn
arXiv 2610.11344sparsity and pruning Published 2026-10-08
ReCal: Calibrating Structured Pruning for On-Policy Distillation Recovery
arXiv 2610.11332quantization Published 2026-10-08
Deflating the Hessian: Rank-4 W4A4 Quantization for Multimodal Diffusion Transformers
arXiv 2610.11315multi-agent orchestration Published 2026-10-08
DuplexAgent-RSI: Recursive Harness Improvement for Full-Duplex Voice Agent Collaboration
arXiv 2610.11299memory and context Published 2026-10-08
REMORY: Learning Residual Memory for Context Compaction
arXiv 2610.11287KV cache Published 2026-10-08
Read What Matters: Query-Adaptive Quantization for KV Caches
arXiv 2610.11245quantization Published 2026-10-08
When Lower Reconstruction Loss Hurts: Distributionally Robust Refinement for Low-Bit LLM Quantization
arXiv 2610.11226safety and guardrails Published 2026-10-08
SafeInferCom: Safe Inference-Time Compute via Verifier-Guided Mid-Generation Intervention for Robotic Task Planning
arXiv 2610.11223efficient architectures Published 2026-10-08
Bridging KV-Cache Quantization and Linear Attention: From Theory to Pretrained Weight Migration
arXiv 2610.11214memory and context Published 2026-10-08
What to Admit and How to Present: Governing Persistent Memory in LLM Agents
arXiv 2610.11188coding agents Published 2026-10-08
Who Pays the Review Cost? Triage, Fairness, and Accountability in AI-authored Pull Requests
arXiv 2610.11179tool use Published 2026-10-08
VAMR: Multi-Question Agentic Reasoning for Efficient Long-Form Video Understanding
arXiv 2610.11171safety and guardrails Published 2026-10-08
Skill Constellations: Tracing the Supply Chain of Agent Skills on GitHub
arXiv 2610.11169efficient training Published 2026-10-08
LadderEdit: Edit-Level Residual Compression for Memory-Efficient Lifelong Editing of LLMs
arXiv 2610.11160memory and context Published 2026-10-08
Do LLMs Learn from Rewards in Context? : Rethinking the role of reward in In-Context Reinforcement Learning
arXiv 2610.11152attention and kernels Published 2026-10-08
QUILT: Rethinking Sparse-Attention Prefill through Shared Query Execution
arXiv 2610.11134multi-agent orchestration Published 2026-10-08
When Interfaces Speak: Data-Aware Generative UI Harness for Active Interaction
arXiv 2610.11123other Published 2026-10-08
Dynamics as Code: On Model Compression via Dynamic System
arXiv 2610.11115agent evaluation Published 2026-10-08
AgentHorizon: Evaluating Agentic Judges for Long-Horizon Computer-Use Tasks
arXiv 2610.11050safety and guardrails Published 2026-10-08
NOMOS: Compiling Written Policies into Statically Verified Tool-Call Gates for LLM Agents
arXiv 2610.11030sparsity and pruning Published 2026-10-07
CircuitTree: Composing Circuits into A Syllable Transcriber
arXiv 2610.12520KV cache Published 2026-10-07
CORAL-KV: Compact Cross-Layer Prediction and Residual Coding for Autoregressive Video KV Caches
arXiv 2610.12514agent evaluation Published 2026-10-07
Prompts versus Rules: Auditing and Controlling Speech Naturalness Behaviors in Voice User Simulators
arXiv 2610.11015memory and context Published 2026-10-07
Curating Always-Loaded Context for LLM Agents: A Capacitated Assortment Model with Censored Feedback
arXiv 2610.11007other Published 2026-10-07
Reconstruction of Multiscale Plasma Dynamics Across Operating Regimes
arXiv 2610.11004other Published 2026-10-07
Low-rank tensor structure of precipitation and its application to satellite-reference merging
arXiv 2610.11000multi-agent orchestration Published 2026-10-07
Cross-Provider Review as a Runtime Contract for Coding Agents: A Controlled Pilot and Fault-Injection Study
arXiv 2610.10961agent evaluation Published 2026-10-07
StoreBench: A Live-Commerce Environment for Evaluating and Training Autonomous Operator Agents
arXiv 2610.10942quantization Published 2026-10-07
Rethinking the Tradeoff Between Temporal Encoding and Nonlinear Computation in Spiking Language Models
arXiv 2610.10933agent evaluation Published 2026-10-07
Reading the Room: Foundations, Design, and Challenges of Normative Competence in LLMs
arXiv 2610.10906multi-agent orchestration Published 2026-10-07
RFChipAgent: Multi-Agentic AI Flow for Analog/RF Chip Design
arXiv 2610.10858agent evaluation Published 2026-10-07
On the Clock: Towards Punctual and Productive Time-Budgeted AI Agents
arXiv 2610.10833failure recovery Published 2026-10-07
Plan-and-Patch: Diffusion Language Models for Agentic Planning
arXiv 2610.10786memory and context Published 2026-10-07
MemoWM: How World Models Change What Agents Need to Remember
arXiv 2610.10778safety and guardrails Published 2026-10-07
BRANCH: Bypassing Multi-Scanner AI Guardrails
arXiv 2610.10742memory and context Published 2026-10-07
Lossy Compressive Text Autoencoders
arXiv 2610.10738efficient architectures Published 2026-10-07
Learning infinite context windows in recurrent architectures via spatial neural computing
arXiv 2610.10690other Published 2026-10-07
Explaining the Saliency Map Sparsity of Adversarially-Trained Neural Networks
arXiv 2610.10666other Published 2026-10-07
Beyond the Ergodic Wall: A Discrete Geometric Physics Sandbox for Analysing AI Scaling Limits and Complexity Collapse
arXiv 2610.10651other Published 2026-10-07
Coverage-Aware Reasoning with Medical Tokens for Diagnosis Prediction
arXiv 2610.10641sparsity and pruning Published 2026-10-07
D-SLR: The Disjoint Row-Sparse plus Low-Rank Decomposition
arXiv 2610.10636efficient training Published 2026-10-07
Phase-HDC: Replacing Optimizer History with Gradient Thresholds in Discrete Phase Learning
arXiv 2610.10630safety and guardrails Published 2026-10-07
The Harness as the Only Mutable Surface: Compliance-Bounded Self-Evolution of LLM Agents in Credit Pipelines, with a Measured Admission Gate
arXiv 2610.10629other Published 2026-10-07
Self-Organization from Constrained Geometric Radiation
arXiv 2610.10621retrieval and RAG Published 2026-10-07
RECAST: Learning to Compute the Right Context through Adaptive Evidence Routing
arXiv 2610.10507other Published 2026-10-07
EmbodiedRSI: Active Continual Robot Learning Through Hypothesis-Guided Co-Evolution
arXiv 2610.10498agent evaluation Published 2026-10-07
Before They Can Solve: Predicting Post-Training Coding-Agent Performance from Base Models
arXiv 2610.10478failure recovery Published 2026-10-07
FoldBack: Self-Correcting Masked Generative Policy for Long-Horizon Garment Folding
arXiv 2610.10462distillation Published 2026-10-07
A Good Self-Teacher Meets the Student Where They Are: Joint On-Policy Learning and Teaching
arXiv 2610.10447memory and context Published 2026-10-07
RunningTab: Direct Workspace Interaction with Environment-Side Tabs
arXiv 2610.10444efficient architectures Published 2026-10-07
Steerspeech: Activation Steering For Emotion Control In Generated Speech
arXiv 2610.10415speculative decoding Published 2026-10-07
Training Parallel Speculative Draft Models by Directly Minimizing Expected Decoding Rounds
arXiv 2610.10411other Published 2026-10-07
SUSpMV: A High Frequency Sparse Matrix Vector Multiplier on HBM Enabled FPGA written in SUS
arXiv 2610.10403quantization Published 2026-10-07
OrBIT: Structure-Guided Embedding Compression
arXiv 2610.10385KV cache Published 2026-10-07
ResidualQuant: KV Cache Quantization for Looped Transformers with 2-Bit Residuals
arXiv 2610.10381efficient training Published 2026-10-07
AutoAdapt: Automatic Domain Discovery Enables Low-Cost Extensibility
arXiv 2610.10349efficient training Published 2026-10-07
Dataset Pruning from First Principles: A Label-Free Linear Programming Approach
arXiv 2610.10347sparsity and pruning Published 2026-10-07
Sparse Planning in Visual World Models via Cost Gradients
arXiv 2610.10274memory and context Published 2026-10-07
Stale, Misattributed, or Late: Where Personal Memory Fails Before Generation
arXiv 2610.10265multi-agent orchestration Published 2026-10-07
When Sub-Agents Work in Parallel: The Promises and Pitfalls of Dynamic Concurrency in Long-Horizon Coding Tasks
arXiv 2610.10263agent evaluation Published 2026-10-07
ProtocolMatch: Protocol-Dependent Model Selection for Scientific Dynamics Forecasting
arXiv 2610.10239coding agents Published 2026-10-07
Why Software Engineering Is Indispensable in the Age of Coding Agents
arXiv 2610.10226failure recovery Published 2026-10-07
Loud Failures, Quiet Failures: Fault Detection and Recovery in Tool-Using Language Model Agents
arXiv 2610.10062memory and context Published 2026-10-07
Cache the Encoder Within:Compact, Reusable Memory across LLM Queries
arXiv 2610.10058memory and context Published 2026-10-07
Learning to Accumulate Knowledge with Mutual Information
arXiv 2610.10042efficient training Published 2026-10-07
Marrying Pricing and Advertising with LLMs
arXiv 2610.09985safety and guardrails Published 2026-10-07
From Expected Harmfulness to Likelihood: A Probabilistic Reformulation of Jailbreaking LLM Agents
arXiv 2610.09973quantization Published 2026-10-07
TR-PTQ: High-Accuracy Integer-Only Transformer Post Training Quantization via Taylor Region Reformulation
arXiv 2610.09969agent evaluation Published 2026-10-07
AgentTime: Can Agents Estimate and Control Their Own Runtime?
arXiv 2610.09944safety and guardrails Published 2026-10-07
AgentTracer: Tracing Indirect Prompt Injection Attack through Fine-Grained Intention-Execution Alignment
arXiv 2610.09935safety and guardrails Published 2026-10-07
Constrained-Action AI Remediation for SIEM/XDR via a NeMo-Guardrails Proxy
arXiv 2610.09906tool use Published 2026-10-07
A Chat Assistant for Software Exploration in a 3D Software Visualization
arXiv 2610.09901quantization Published 2026-10-07
Layerwise Error Attribution for Fast and Robust Mixed-Precision Post-Training Quantization
arXiv 2610.09877agent evaluation Published 2026-10-07
LiveMACE: Process-Aware Evaluation of LLM Agent Capabilities in Evolving Markets
arXiv 2610.09872failure recovery Published 2026-10-07
Training Advisors for LLM Agents from Task Outcomes
arXiv 2610.09858efficient training Published 2026-10-07
ORCA: Hunting Compositional Failures in Text-to-Image Diffusion
arXiv 2610.09841memory and context Published 2026-10-07
SkillForge: Co-Evolving Skills and Agents via Dynamic Skill Lifecycles
arXiv 2610.09832KV cache Published 2026-10-07
Dual-QK: Sharp Queries and Flat Keys for Prunable 2-bit KV Caches
arXiv 2610.09827other Published 2026-10-07
DisParQ: Self-Supervised Part Concepts for Interpretable Vision Foundation Models
arXiv 2610.09802other Published 2026-10-07
AdaPS-LiNGAM: Adaptive Predecessor Selection for Linear Non-Gaussian Acyclic Models under Small-Sample Settings
arXiv 2610.09782retrieval and RAG Published 2026-10-07
EntroPrefill: Renyi-Guided Context Pruning with Conditional Stability Guarantees for Retrieval-Augmented Generation
arXiv 2610.09757efficient architectures Published 2026-10-07
SpikingVLA: Asynchronous Spiking Vision-Language-Action Models
arXiv 2610.09710safety and guardrails Published 2026-10-07
Understanding and Mitigating Token-Pruning-Induced Vulnerabilities in VLMs
arXiv 2610.09703other Published 2026-10-07
The Silhouette Operator: Identifiability of Low-Rank Measures from One-Dimensional Projections
arXiv 2610.09687multi-agent orchestration Published 2026-10-07
Cost-Efficient Theorem Proving via Agent Orchestration in Program Verification
arXiv 2610.09681other Published 2026-10-07
Physics-Informed Neural Plasticity: PDE Solvers That Reshape Themselves
arXiv 2610.09510KV cache Published 2026-10-07
CHASE: Channel-Aligned Structure Exploitation for Geometry-Aware Model Engineering
arXiv 2610.09476other Published 2026-10-07
DSReg: Provably Recovering Individual World Latents without Reconstruction
arXiv 2610.09457serving and batching Published 2026-10-07
Democratizing MoE inference on commodity GPUs with CoMoE
arXiv 2610.09424quantization Published 2026-10-07
OnlineQAT: On-Policy Distillation for Ultra-Low-Bit Large Language Models
arXiv 2610.09346sparsity and pruning Published 2026-10-07
Shared Low-rank Basis Factorization for Data-free Mixture-of-Experts Compression
arXiv 2610.09342planning and decomposition Published 2026-10-07
SearchWorld: Spatial Value-Grounded Imagination for UAV Object Search via World Models
arXiv 2610.09335other Published 2026-10-07
An Interleaved Parallel Dependent Quantization Hardware Architecture for H.266/VVC
arXiv 2610.09292quantization Published 2026-10-06
Quantize by Drift: Label-Free Mixed-Precision Post-Training Quantization for Text Embedders
arXiv 2610.09227quantization Published 2026-10-06
CurveTQ: Rotation-Free Trellis Quantization of LLM Weights via Curvature-Weighted Search
arXiv 2610.09212quantization Published 2026-10-06
Few Bits, One Law: Toward W2A4KV2
arXiv 2610.09202efficient training Published 2026-10-06
Exact Dynamics and Finite-Sample Trajectory Recovery of Linear Recursive Feature Machines
arXiv 2610.09196quantization Published 2026-10-06
Q-PACE: Dynamic Precision Allocation for Quantization-Aware Training
arXiv 2610.09183efficient training Published 2026-10-06
MaRK: Markov-adapted Recurrent Kernels for Dynamic Operator Conditioning in State Space Models
arXiv 2610.09092attention and kernels Published 2026-10-06
Tucker Bottleneck Attention for Multi-Dimensional Sequence Modeling
arXiv 2610.09090sparsity and pruning Published 2026-10-06
TAP: Efficient Long-Horizon Agent Pruning via Trajectory-Anchored Recovery
arXiv 2610.09074distillation Published 2026-10-06
Multi-Label Topic Assignment via LLM Distillation: A Comparative Analysis of Generative vs. Discriminative Student Models
arXiv 2610.09063KV cache Published 2026-10-06
A Self-Pruning Transformer: Extreme KV-Cache Compression with Universal Attention
arXiv 2610.09051planning and decomposition Published 2026-10-06
Shared-Roadmap Generation and Evaluator for Multi-Agent Path Planning Using Heterogeneous Graph Neural Network
arXiv 2610.09034attention and kernels Published 2026-10-06
SPIN: Shadow Predictive Indexer for Sparse Attention
arXiv 2610.09025safety and guardrails Published 2026-10-06
How Fragile Is On-Device Language Model Safety? Localizing Safety-Critical Parameters for Sparse Fault Analysis
arXiv 2610.09000other Published 2026-10-06
SNR-Gated LSTM-Conditioned Diffusion Model for MIMO Channel Estimation
arXiv 2610.08977agent evaluation Published 2026-10-06
CARE: Certifying Acceleration for Vision-Language-Action Inference
arXiv 2610.08917quantization Published 2026-10-06
Task-Sufficient Contraction: Source Selection for Machine Information Interfaces
arXiv 2610.08884other Published 2026-10-06
Reinforcement Learning with Conformal Action Sets: An Application to Sequential Recommendation
arXiv 2610.08743retrieval and RAG Published 2026-10-06
A Systematic Study of Semantic ID Spaces for Generative Information Retrieval
arXiv 2610.08732speculative decoding Published 2026-10-06
Secure Speculative Decoding for Large Language Models
arXiv 2610.08678efficient training Published 2026-10-06
MemFLoRA: Memory-Floor LoRA for CNN Adaptation at the Edge
arXiv 2610.08669attention and kernels Published 2026-10-06
Random Feature Gaussian Process Attention: Linear-Time Probabilistic Attention with Calibrated Uncertainty
arXiv 2610.08578other Published 2026-10-06
FlowCF: Sparse Counterfactual Explanations for Mixed-Type Tabular Data using Flow Matching
arXiv 2610.08537quantization Published 2026-10-06
ActTune: Action-Aware Precision and GPU Operating-Point Adaptation for Energy-Efficient Vision-Language-Action Inference
arXiv 2610.08444efficient training Published 2026-10-06
NeMo-DCR: Bit-Exact Delta-Compressed Refit for Scalable Agentic RL at Trillion-Parameter Scale
arXiv 2610.08430quantization Published 2026-10-06
SSR: Sparse Segment Reduction for Ternary GEMM Acceleration
arXiv 2610.08403retrieval and RAG Published 2026-10-06
Foresight-over-Graph: Reasoning Beyond Local Horizons for Knowledge Base Question Answering
arXiv 2610.08388KV cache Published 2026-10-06
Lachesis: Lifetime-Aware KV Cache Placement for Agent Serving across HBM and High-Bandwidth Flash
arXiv 2610.08378other Published 2026-10-06
Evolutionary One-Step Generators: Fast and Diverse Sampling for Discrete Design
arXiv 2610.08367sparsity and pruning Published 2026-10-06
DIPrune: Task-Aware Token Pruning with Dual Importance for Efficient Multimodal Language Models
arXiv 2610.08341efficient training Published 2026-10-06
CoDe-LoRA: Mitigating the Orthogonality Dilemma in Continual Learning of LLMs via Knowledge Consolidation and Decoupling
arXiv 2610.08312serving and batching Published 2026-10-06
DySCo: Dynamic Sharding for Collaborative Edge-Cloud LLM Inference with Depth-Synchronized Batching
arXiv 2610.08268safety and guardrails Published 2026-10-06
HE-OFT: Privacy-Preserving One-Shot Federated Fine-Tuning under Homomorphic Encryption
arXiv 2610.08255quantization Published 2026-10-06
Align, Then Correct: Training-Free Two-Stage Low-Rank Compensation for Extremely Quantized Large Language Models
arXiv 2610.08164safety and guardrails Published 2026-10-06
POLAR: Ontology-Guided Risk Prevention for Tool-Calling LLM Agents
arXiv 2610.08082efficient training Published 2026-10-06
A Riemannian Geometry for Low-rank Adaptation
arXiv 2610.08049sparsity and pruning Published 2026-10-06
VisionWeave: Weaving Elastic Visual Representations as a Native Capability of MLLMs
arXiv 2610.07987KV cache Published 2026-10-06
Hybrid Latent Attention for Looped Language Models
arXiv 2610.07940quantization Published 2026-10-06
ApexQuant: Data-Free Elastic Quantization by Residual Re-Isotropization
arXiv 2610.07904memory and context Published 2026-10-06
ReFold: Training-Free Reversible Inter-Turn Context Folding for Long-Horizon Agents
arXiv 2610.07863efficient training Published 2026-10-06
Tram-FL: Reducing Communication and Computation Costs through Sequential Model Circulation in Decentralized Federated Learning
arXiv 2610.07859quantization Published 2026-10-06
Lost in the bf16 Cast: Exporting Ternary Language Models Can Revert Most Low-Learning-Rate Code Changes
arXiv 2610.07853speculative decoding Published 2026-10-06
Nucleus Speculative Decoding: Plausibility-Aware Verification Beyond Exact Distribution
arXiv 2610.07822tool use Published 2026-10-06
Do I Need the Cloud? Uncertainty-Aware Step-Level Handoff for Small Language Model Agents
arXiv 2610.07816agent evaluation Published 2026-10-06
Persistent Memory in Multi-Agent LLM Inference: What It Costs, What It Buys, and When You Can Tell
arXiv 2610.07782agent evaluation Published 2026-10-06
Quantization Effects on Tool-Failure Recovery Vary Across Prompts and Evaluation Designs
arXiv 2610.07781speculative decoding Published 2026-10-06
APEX: Speculate smarter, not deeper
arXiv 2610.07780quantization Published 2026-10-06
TRACE: Rollout-Guided Quantization-Aware Training for FP4 Reinforcement Learning of MoE Language Models
arXiv 2610.07767efficient architectures Published 2026-10-06
SanSi: A Looped Typed Decision Model for System 1.5 Thinking
arXiv 2610.07730memory and context Published 2026-10-05
Recursive Video In-Context Learning for Agentic Robot
arXiv 2610.06843efficient architectures Published 2026-10-05
Towards Looped Models Done Right, Part II: Rethinking at Fixed Points
arXiv 2610.06833memory and context Published 2026-10-05
MemPilot: Orchestrating On-Demand Multimodal Memory Curation for LLM Agents
arXiv 2610.06830agent evaluation Published 2026-10-05
PlotGround: Grounding Plot Digitization in Real Scientific Figures and Their Source Data
arXiv 2610.06825distillation Published 2026-10-05
Paradee: Distilling Kokoro-82M into an 8M-Parameter Single-Voice Text-to-Speech Model
arXiv 2610.06817multi-agent orchestration Published 2026-10-05
Back to the Future: Rethinking EDA Infrastructure for Agentic Systems in Chip Design Verification
arXiv 2610.06790other Published 2026-10-05
MatrixFormer: A Foundation Model for Matrix Completion
arXiv 2610.06751agent evaluation Published 2026-10-05
BazaarBench: Delegation Safety in Decentralized C2C Marketplaces Run by LLM Agents
arXiv 2610.06748agent evaluation Published 2026-10-05
ufakzeka-karar: An Open Turkish Typed-Decision Model with Order-Invariant Option Scoring
arXiv 2610.06744multi-agent orchestration Published 2026-10-05
MedPrune: Topology-Efficient Multimodal Multi-Agent Communication Evolution for Medical VQA Tasks
arXiv 2610.06695KV cache Published 2026-10-05
OVAL: Output-Aware Local Page Bases for KV Cache Retrieval
arXiv 2610.06686efficient architectures Published 2026-10-05
How Sparse Probability Maps Shape Mixture-of-Experts Routing
arXiv 2610.06677distillation Published 2026-10-05
Learning What to Imitate: Entropy-Aware Distribution Mixing
arXiv 2610.06671memory and context Published 2026-10-05
Wikidata Search Traces: A Dataset for Training Knowledge Graph Search Agents
arXiv 2610.06650efficient training Published 2026-10-05
LoGRA: Scaling LLM Reinforcement Learning with Low-Rank Gradient Sketches
arXiv 2610.06647efficient training Published 2026-10-05
Frozen Factor or Spectral Band? Disentangling Two Choices in Low-Rank LoRA
arXiv 2610.06621other Published 2026-10-05
RealtimeWAM: One-Step Asynchronous World Action Models
arXiv 2610.06617serving and batching Published 2026-10-05
Can Agent Harnesses and Inference Engines Hear Each Other? The HEAR Protocol for Agentic LLM Serving
arXiv 2610.06597agent evaluation Published 2026-10-05
Mind the Accent Gap: British Accent Robustness in Speech-Driven Financial Voice Assistants
arXiv 2610.06587safety and guardrails Published 2026-10-05
Before Agent Tells The Lie: Has Deception Already Been Represented?
arXiv 2610.06576failure recovery Published 2026-10-05
HERA: Harness-Environment Co-Evolution for Reliable Agentic Abstention
arXiv 2610.06563other Published 2026-10-05
Empirical Variational Autoencoder
arXiv 2610.06545efficient training Published 2026-10-05
A Fine-Grained Analysis of the LoRA Fine-Tuning Landscape with Implications for Data Selection
arXiv 2610.06542safety and guardrails Published 2026-10-05
ANT: A Multi-Granularity Network Traffic Dataset and Benchmark for Agents Behavior Auditing
arXiv 2610.06514KV cache Published 2026-10-05
Behavior-Preserving KV Cache Compression
arXiv 2610.06479safety and guardrails Published 2026-10-05
AgentPrivArena: Evaluating and Auditing Real-world AI Agent Privacy
arXiv 2610.06454safety and guardrails Published 2026-10-05
Efficient Secure Federated Learning via Information-Theoretically Secure Key Distribution: A Medical Imaging Case Study
arXiv 2610.06420efficient training Published 2026-10-05
Quantifying the Stability of Multi-Step Reasoning via Error Amplification
arXiv 2610.06404safety and guardrails Published 2026-10-05
RAISED: Self-Distillation for Robustness to Prompt Injection in LLM Agents
arXiv 2610.06401efficient training Published 2026-10-05
Scaling Down the Scaling Laws: Parameter Efficiency and Compute-Optimal Training in Resource-Constrained Large Language Models
arXiv 2610.06387KV cache Published 2026-10-05
DeferKV: Rethinking Eviction Timing for One-Shot KV Cache Compression
arXiv 2610.06286quantization Published 2026-10-05
Shared Stopping Decisions Change Answers in HQQ Cache Quantization
arXiv 2610.06251efficient training Published 2026-10-05
Do Small Language Models Learn to Negotiate? A Controlled Scaling Study of RL-Trained Sellers
arXiv 2610.06204agent evaluation Published 2026-10-05
Correct Code, Broken Contributions? SWE-CC: Benchmarking Repository Policy Compliance for Coding Agents
arXiv 2610.06193multi-agent orchestration Published 2026-10-05
Copies or Sources? Measuring How LLM Aggregators Count Restated Evidence in Multi-Agent Systems
arXiv 2610.06192safety and guardrails Published 2026-10-05
Anosognosia in LLMs: Probing Self-Awareness of Quantized Computational Substrate
arXiv 2610.06174safety and guardrails Published 2026-10-05
Benchmarking Jailbreak Guardrails for Embodied Agents
arXiv 2610.06122agent evaluation Published 2026-10-05
From Traces to Agentic Worlds: Agentic Language World Models for Interactive Environment Simulation
arXiv 2610.06100safety and guardrails Published 2026-10-05
TrustMI: Causally controlling how assistants trust their users
arXiv 2610.06064planning and decomposition Published 2026-10-05
RocketAgent: A Long-Horizon Engineering Agent for Multidisciplinary Design of Liquid-Rocket Thrust Chambers
arXiv 2610.06044distillation Published 2026-10-02
Language Models that Play Chess and Explain Their Moves
arXiv 2610.03695tool use Published 2026-10-02
From Benchmarks to Production: A Text-to-SQL System for Complex Financial Data
arXiv 2610.03524safety and guardrails Published 2026-10-02
Passing the Test You Trained On: Re-evaluating Prompt-Injection Detectors for LLM Agents
arXiv 2610.03448quantization Published 2026-10-02
16-bit Precision of Convolutional Neural Networks on Microcontroller Units for 8-bit Costs
arXiv 2610.03402multi-agent orchestration Published 2026-10-02
EdgeAgent: Orchestrating On-Device LLM inference for End-User Multi-Agent Systems on CPU-GPU Unified Memory Architectures
arXiv 2610.03394sparsity and pruning Published 2026-10-02
From Patching to Pruning Visual Computation in Vision Language Models
arXiv 2610.03389agent evaluation Published 2026-10-02
ReFract: Benchmarking Perspective Awareness in Language Model Agents with Text World Models
arXiv 2610.03356agent evaluation Published 2026-10-02
Lightweight, Rubric-Guided Trajectory Evaluation for Production AI Agents
arXiv 2610.03315other Published 2026-10-02
S$^{2}$-PINN: Stochastic Separable Physics-Informed Neural Networks
arXiv 2610.03303efficient training Published 2026-10-02
VenusRL: A Fully Disaggregated Agentic RL System with Priority Scheduling and Scalable Interaction
arXiv 2610.03286coding agents Published 2026-10-02
D2K-Bench: Can LLM Agents Turn Expert Designs into Efficient GPU Kernels?
arXiv 2610.03226efficient training Published 2026-10-02
AdaStep: Adaptive Step Credit Weighting for Agentic Reinforcement Learning
arXiv 2610.03223safety and guardrails Published 2026-10-02
Toward SLM-based agentic task-tool intent matching
arXiv 2610.03213KV cache Published 2026-10-02
KV$^2$: A Self-Refining KV Cache
arXiv 2610.03198agent evaluation Published 2026-10-02
Source Preference in the Wild: How LLM Agents Favor Items by Source, and How to Reduce It
arXiv 2610.03195multi-agent orchestration Published 2026-10-02
FinNextAssist: Towards Professional Financial Deep Research Assistant
arXiv 2610.03174KV cache Published 2026-10-02
Page-EntroKV: Hardware-Aligned, Entropy-Weighted KV-Cache Eviction under Grouped-Query Attention
arXiv 2610.03135efficient architectures Published 2026-10-02
Coverage You Can Steer: Online Conformal Calibration for RL-Driven Hardware-Aware NAS
arXiv 2610.03127sparsity and pruning Published 2026-10-02
Exploring the Trade-Off Between Structured Pruning and Fault Tolerance in Deep Neural Networks for Space Applications
arXiv 2610.03117KV cache Published 2026-10-02
Emergent Structure in the Marginal Attention Space of Language Models
arXiv 2610.03109agent evaluation Published 2026-10-02
Ask, Relax, or Act? Evaluating Actionable Indeterminacy in LLM Preference Reasoning
arXiv 2610.03102safety and guardrails Published 2026-10-02
Securing Computer-Use Agents Against Branch Steering Attacks
arXiv 2610.03089multi-agent orchestration Published 2026-10-02
MOF-VERIFY: A Failure-Aware Agentic Harness for MOF Hypothesis Verification
arXiv 2610.03056safety and guardrails Published 2026-10-02
hacktrace: behavior-supervised detection of reward hacking during code generation
arXiv 2610.03055distillation Published 2026-10-02
HyperThink: Text-to-Parameter Hypernetworks for Efficient Reasoning
arXiv 2610.03039failure recovery Published 2026-10-02
WebFovea: When the Model Is Right but the Click Is Wrong -- Reliable Round Trips for Vision-Based Web Agents on Live Websites
arXiv 2610.03036KV cache Published 2026-10-02
Tailoring the Quantization Space for 1-Bit KV Cache Compression
arXiv 2610.03027agent evaluation Published 2026-10-02
DyadMem: A Long-Term Memory Benchmark of How Agents Work with Users
arXiv 2610.03020safety and guardrails Published 2026-10-02
Beyond Predefined Sinks: Security-Aware Dependency Analysis for LLM Agents
arXiv 2610.03014KV cache Published 2026-10-02
AvoKV-E: Payload-Aware KV Cache Eviction for Long Reasoning
arXiv 2610.03007failure recovery Published 2026-10-02
Sentry: Learning to Recover from LLM Agent Failures at Test Time
arXiv 2610.02994KV cache Published 2026-10-02
SlimKV: Joint Token-Feature KV Cache Compression with Reconstruction-Free Beacon Attention
arXiv 2610.02953coding agents Published 2026-10-02
GTDD: Generative Test-Driven Development for AI Coding Agents with Adversarial Testing
arXiv 2610.02952sparsity and pruning Published 2026-10-02
Dynamic Expert Pruning for Multi-Agent Systems
arXiv 2610.02951memory and context Published 2026-10-02
Continual Graph Memory for Mathematical Research Agents
arXiv 2610.02945safety and guardrails Published 2026-10-02
HASTE: Evolving Agent Harnesses Against Emerging Attacks Using Sparse Evidence
arXiv 2610.02920other Published 2026-10-02
Do ResNets Route? Sparse Interaction Experts in Residual Networks
arXiv 2610.02907efficient architectures Published 2026-10-02
DyRA: Dynamic Residual Approximation for Efficient Matrix Multiplication in DNNs
arXiv 2610.02882retrieval and RAG Published 2026-10-02
Query-aware routing for Cross-lingual performance gains in Encoders
arXiv 2610.02875efficient training Published 2026-10-02
ByteSplat: Efficient Distributed 3D Gaussian Splatting Training via Intra- and Inter-GPU communication reduction
arXiv 2610.02851tool use Published 2026-10-01
KaliBench: A Fine-Grained Benchmark for Cybersecurity Tool Use on Kali Linux with Runtime-Free Verifiable Rewards
arXiv 2610.02206memory and context Published 2026-10-01
AutoCompact: Learning When to Compact Context in Long-Horizon Coding Agents
arXiv 2610.02163memory and context Published 2026-10-01
From Knowledge Access to Source Learning: Developing Source-Specific Competence
arXiv 2610.02150agent evaluation Published 2026-10-01
Keyword Harnesses Fail Open: A Cheap Diagnostic Ladder for Tool-Use Claims in Small Language Models
arXiv 2610.02142tool use Published 2026-10-01
HumanoidToolBench: Benchmarking Humanoid Tool Use from Selection to Mobile Execution
arXiv 2610.02089efficient training Published 2026-10-01
Learn the Directions, Normalize the Gains: Post-Training Normalization for LoRA
arXiv 2610.02067tool use Published 2026-10-01
Mimir: Physics-Grounded LLM Agents for Long-Horizon Irrigation Control
arXiv 2610.02038memory and context Published 2026-10-01
Mem++: Non-Destructive Memory for Long-Term Organizational LLM Agents
arXiv 2610.02002failure recovery Published 2026-10-01
Mingbird: A Local-First Agent Harness Enabling Small Open Models to Complete Real Tasks
arXiv 2610.02001other Published 2026-10-01
FastCI: Efficient GPU-Intensive CI for LLM Training Frameworks
arXiv 2610.01967planning and decomposition Published 2026-10-01
Selection-Based Structured Reasoning: Toward Efficient Multimodal Search Agents
arXiv 2610.01892agent evaluation Published 2026-10-01
Continuous Process-Level Evaluation for Evolving Enterprise AI Agent Skills
arXiv 2610.01833coding agents Published 2026-10-01
CONTRA: Discovering and Qualifying Behavior-Changing Questions for Selective Clarification in LLM Code Generation
arXiv 2610.01769agent evaluation Published 2026-10-01
VideoEvolve: Evolving Agent Harnesses for Video Temporal Grounding
arXiv 2610.01766efficient training Published 2026-10-01
Task-Oriented Rank Adaptation for Continual Learning in Text Classification
arXiv 2610.01702other Published 2026-10-01
MiLoop: Selective Memory Propagation for Neural Combinatorial Optimization
arXiv 2610.01685sparsity and pruning Published 2026-10-01
FedSAP: Federated Learning with Structured Adaptive Partitioning for Multi-Domain Heterogeneous Edge Devices
arXiv 2610.01638agent evaluation Published 2026-10-01
Agents Are Systems, Not Models: Rethinking Agentic Evaluation
arXiv 2610.01618other Published 2026-10-01
Exploiting the Interplay of Compute- and Memory-Bound kernels in MPI Applications
arXiv 2610.01587safety and guardrails Published 2026-10-01
Chaining Skills to Hijack LLM Agents
arXiv 2610.01564sparsity and pruning Published 2026-10-01
QK-Wanda: Coupling Queries and Keys for Unstructured Pruning
arXiv 2610.01554efficient training Published 2026-10-01
FedFit: Federated Fine-Tuning of LLMs via Vector-Bank Parameterization and Quantization
arXiv 2610.01537efficient training Published 2026-10-01
Sharpening Tax in Post-Training
arXiv 2610.01509safety and guardrails Published 2026-10-01
OverAct: Measuring and Mitigating Proactive Over-Authorization in LLM Tool-Calling Agents
arXiv 2610.01508failure recovery Published 2026-10-01
The Persona Is Still There, but Who Is Speaking? Latent Identity Reversion in Persistent AI Agents
arXiv 2610.01490sparsity and pruning Published 2026-10-01
MWOP: Modality-aware Width-wise Operation Pruning for Efficient MLLMs
arXiv 2610.01434memory and context Published 2026-10-01
Beyond Memory: Harnessing Long-Horizon Agents with Explicit Belief States
arXiv 2610.01415safety and guardrails Published 2026-10-01
PACE: Provenance-Aware Capability Enforcement for Tool-Using LLM Agents
arXiv 2610.01349other Published 2026-10-01
An ontology for cross-sectoral crisis management: core and public health modules
arXiv 2610.01326efficient training Published 2026-10-01
Know When to Hold 'em: Correct-Token Retention in Uniform-State Diffusion Language Models
arXiv 2610.01275multi-agent orchestration Published 2026-10-01
Science Utopia? Closed-Loop LLM Simulation of Academic Research Ecosystems
arXiv 2610.01257failure recovery Published 2026-10-01
DeFA: Dependency-Guided Failure Attribution for LLM Agents
arXiv 2610.01256agent evaluation Published 2026-10-01
AGO AI Quality Gate: Evidence-First Release Decisions for Retrieval-Augmented Generation
arXiv 2610.01218multi-agent orchestration Published 2026-10-01
Federated Agent Optimization
arXiv 2610.01195safety and guardrails Published 2026-10-01
HeadEdit: Calibrating Language Model Behavior Through the Frozen Unembedding Matrix
arXiv 2610.01170tool use Published 2026-10-01
CineMR: Tool-Integrated Vision-Language Reasoning for Quantitative Cardiac MRI Assessment
arXiv 2610.01166retrieval and RAG Published 2026-10-01
Do Multilingual Encoders Produce Language-Consistent Semantic IDs?
arXiv 2610.01139multi-agent orchestration Published 2026-10-01
Auditing Action Settlement in LLM Agent Environments: Order, Progress, and Replay
arXiv 2610.01138speculative decoding Published 2026-10-01
AgSpec: Pushing the Limits of Retrieval-Based Speculative Decoding in Coding Agent Pipelines
arXiv 2610.01108other Published 2026-10-01
Polylogarithmic Sparsity of Randomly Reweighted NPMLEs for Gaussian Mixtures
arXiv 2610.01088coding agents Published 2026-10-01
Network World Models as Environments for Algorithm Design on Complex Systems
arXiv 2610.01048agent evaluation Published 2026-10-01
Groundability, Not Scale Alone: When Weak Reviewers Can Audit Strong Coding Agents
arXiv 2610.01023quantization Published 2026-10-01
The Devil Is in the Reconstruction Loss Scale: Rethinking Optimization in LLM Quantization
arXiv 2610.00983multi-agent orchestration Published 2026-10-01
Can AI Scientists Coordinate at Runtime?
arXiv 2610.00980efficient training Published 2026-10-01
RISED: RubrIcs for agentic multi-environment Selection and sElf-Distillation
arXiv 2610.00979safety and guardrails Published 2026-10-01
ABSENTIA: Detecting Broken Access Control Vulnerabilities in Web Applications
arXiv 2610.00977agent evaluation Published 2026-10-01
VeriHarness: Scaling Agentic Verification for Long-Horizon Tasks
arXiv 2610.00972memory and context Published 2026-10-01
RPTune: Learned Context Curation for LLM Catalog Search
arXiv 2610.00964memory and context Published 2026-10-01
Role-aware Heuristic Episodic Attention for Conversational LLMs
arXiv 2610.00958efficient training Published 2026-10-01
PG-SFT: Balancing Capability Acquisition and Retention in Offline Agent Fine-Tuning
arXiv 2610.00949efficient training Published 2026-10-01
Platonic Task Arithmetic
arXiv 2610.00929agent evaluation Published 2026-10-01
Finding the Right Fit: Model-Harness Interactions across Agent Tasks
arXiv 2610.00917other Published 2026-10-01
The Geometry of Contextual Relations: Language Models Address Facts by Order of Mention
arXiv 2610.00910failure recovery Published 2026-10-01
ActiveSaddler: Automated Curriculum Learning for Agent Harness Optimization
arXiv 2610.00906other Published 2026-10-01
TOAST: Stochastic Robot Action Tokenization for Autoregressive Vision-Language-Action Models
arXiv 2610.00899efficient architectures Published 2026-10-01
Clock Diffusion: Efficient Semi-Autoregressive Continuous Diffusion Language Models
arXiv 2610.00894efficient training Published 2026-10-01
Cross-Benchmark Transfer from RL on Agentic Coding Tasks
arXiv 2610.00890safety and guardrails Published 2026-10-01
FORALL-LEAN-AGENT for Auditable Reasoning in Formal Mathematics and Software Verification
arXiv 2610.00885memory and context Published 2026-10-01
MemFit: Efficient Long-Term Agentic Memory
arXiv 2610.00872other Published 2026-10-01
Kinematic MeanFlow: One-Step Action Generation Policy for Robotic Foundation Models
arXiv 2610.00864safety and guardrails Published 2026-10-01
AuraForge: Scaling Security Supervision for Training Coding Agents
arXiv 2610.00850safety and guardrails Published 2026-09-30
Sapien: A Stateful Policy Engine for Autonomous AI Agents
arXiv 2610.00797agent evaluation Published 2026-09-30
Scalable Multi-Task Inverse Reinforcement Learning
arXiv 2610.00758quantization Published 2026-09-30
Q-MINO: A Minimal-Norm Method for Quantization-Aware Training
arXiv 2610.00738other Published 2026-09-30
Benchmarking Generative Models for Weather Data Assimilation on Real Station Observations
arXiv 2610.00728agent evaluation Published 2026-09-30
ReLiveGym: Evaluating Long-Lived Agents over Weeks of Replayed Reality
arXiv 2610.00710quantization Published 2026-09-30
How Divergence Becomes Decision Flips in Compressed Language Models
arXiv 2610.00694memory and context Published 2026-09-30
LabBook: Harnessing Experimental History for Efficient LLM-Driven Discovery
arXiv 2610.00675quantization Published 2026-09-30
Analysis of Quantized and Efficiently Adapted Protein Language Models
arXiv 2610.00665efficient training Published 2026-09-30
PhysicsMate: A Curriculum-Grounded Bengali Benchmark for Secondary Physics QA with Small-Model Adaptation
arXiv 2610.00664agent evaluation Published 2026-09-30
Agent Evaluation Reliability: More Tasks Won't (Always) Fix An Agent Leaderboard
arXiv 2610.00651coding agents Published 2026-09-30
Self-Evolving Coding Rules for AI Coding Agents
arXiv 2610.00650agent evaluation Published 2026-09-30
Incident-Arena: Getting agents to the last nine of reliability
arXiv 2610.00648efficient training Published 2026-09-30
Explainable Suicide Risk Assessment on Social Media with Multi-Task QLoRA
arXiv 2610.00610agent evaluation Published 2026-09-30
Legal Research Bench: Measuring End-to-End Reliability in Long-Horizon Legal Research Agents
arXiv 2610.00609efficient training Published 2026-09-30
Make Sparse Rewards Count: Density-Aware Reward Aggregation for Multi-Reward RL
arXiv 2610.00574other Published 2026-09-30
Scaling Collider Event Generation with Residual-Quantized Tokens
arXiv 2610.00569attention and kernels Published 2026-09-30
Attention Kernels for Learning Maps Between Heavy-Tailed Measures
arXiv 2610.00564agent evaluation Published 2026-09-30
No One Architecture Fits All: A Cross-Environment Evaluation of Hierarchical Red Team Agents
arXiv 2610.00557efficient training Published 2026-09-30
One-Step Generative Modeling via Training Dynamics Action
arXiv 2610.00518other Published 2026-09-30
Heteroskedastic Canonical Polyadic Tensor Decomposition
arXiv 2610.00498efficient training Published 2026-09-30
Score the Update, Not the Token: Descent-Aligned Routing for Combinatorial LoRA Experts
arXiv 2610.00493other Published 2026-09-30
AIR-LLM: Broadcasting AI Weights over Radio for Memory-Free Edge LLM Inference via RF Computing
arXiv 2610.00465planning and decomposition Published 2026-09-30
JevSpawn: Adaptive Agentic Inference through Compositional Action Spaces
arXiv 2610.00437quantization Published 2026-09-30
XOR-Trellis: Ultra-Low-Complexity Dequantization and Curvature-Aware Hadamard-Free LLM Quantization
arXiv 2610.00432safety and guardrails Published 2026-09-30
Memetic Trojans: Social Contagions as Carriers of Adversarial Payloads in Agent Networks
arXiv 2610.00430sparsity and pruning Published 2026-09-30
IrekoGPT: Turning Structured Pruning into Post-Hoc Slimmable LLMs
arXiv 2610.00426agent evaluation Published 2026-09-30
Code That Works, Environments That Don't: Measuring Environment Reproducibility in AI-Generated Software
arXiv 2610.00425KV cache Published 2026-09-30
EchoPress: Query-Agnostic KV Cache Pruning via Virtual Context Reconstruction
arXiv 2610.00412efficient training Published 2026-09-30
T2SPO: Trajectory-to-Step Policy Optimization for Agentic Reinforcement Learning
arXiv 2610.00388quantization Published 2026-09-30
On the Relationship between Model Quantization and Model Inversion Attacks
arXiv 2610.00382sparsity and pruning Published 2026-09-30
MoRA: MoE Pruning via Router Bias Learning and Expert Approximation
arXiv 2610.00367multi-agent orchestration Published 2026-09-30
Cogentic: Multi-Agent Orchestration for Automated Proof Discovery
arXiv 2609.40324efficient architectures Published 2026-09-30
Scaling Laws for Looped Mixture of Experts
arXiv 2609.40316safety and guardrails Published 2026-09-30
Compression Footprints as Security Signals for Model-Poisoning Defense in Federated Learning
arXiv 2609.40312multi-agent orchestration Published 2026-09-30
How Much of a Harness Does a Strong Agent Need for Autonomous ML Engineering?
arXiv 2609.40303agent evaluation Published 2026-09-30
cua-speedrun: Standardized Benchmarking of the Speed of Computer-Use Agents
arXiv 2609.40284other Published 2026-09-30
OpenTSLM TeeMoE: A Unified Time-Series Language Model for Forecasting, Contextual Prediction, and Reasoning
arXiv 2609.40265distillation Published 2026-09-30
ComputerSD: Online Self-Distillation from Real-Time Feedback for Computer-Use Agents
arXiv 2609.40253efficient training Published 2026-09-30
PhantomEnvironments: Training LLM Agents in Fictional Worlds
arXiv 2609.40221agent evaluation Published 2026-09-30
SCB: SpeechConversationBench for Evaluating Multi-Turn Reasoning in Speech-to-Speech Models
arXiv 2609.40198tool use Published 2026-09-30
Learning from Research: Toward Lifelong Agent Harness Evolution
arXiv 2609.40169other Published 2026-09-30
Reinforcement Learning-Guided Graph Transformations for SpTRSV Optimization
arXiv 2609.40159efficient architectures Published 2026-09-30
Learning Functional Subspaces for Neural Network Compression
arXiv 2609.40127memory and context Published 2026-09-30
Persistent Context Graphs for Efficient Memory Compaction in LLM Agents
arXiv 2609.40118failure recovery Published 2026-09-30
Agent Error Dataset: Scaling 50,000 Error--Diagnosis Pairs for Failure Analysis and Error-Aware Post-Training
arXiv 2609.40111agent evaluation Published 2026-09-30
JuryFlow: Disagreement-Guided Human-in-the-Loop Multi-Agent Evaluation
arXiv 2609.40103agent evaluation Published 2026-09-30
AutoDataBench: A Data-centric Testbed for Accelerating Auto Research
arXiv 2609.40097distillation Published 2026-09-30
MeanVoiceFlow2: Joint Optimization of Mean Flow and Content Encoder for Fast One-Step Zero-Shot Voice Conversion
arXiv 2609.40087other Published 2026-09-30
Accelerated Algorithm for Sparse Regularized Partial Optimal Transport
arXiv 2609.40075efficient training Published 2026-09-30
LARC: Low-Rank Adaptive Residual Connections for Learning in Frozen Models
arXiv 2609.40063safety and guardrails Published 2026-09-30
Who Verifies the Graph? Misspecification Attacks on Causal Action Verification for Language Agents
arXiv 2609.40027coding agents Published 2026-09-30
Mid-Harness: Scaling Actions Between Model and Harness for Terminal Agents
arXiv 2609.39982speculative decoding Published 2026-09-30
UBTree: Parallel Tree Drafting via Unigram and Bigram Models for Speculative Decoding
arXiv 2609.39972agent evaluation Published 2026-09-30
Better Deck or Different Judge? Evaluating Agentic Harness Gains in Corporate and Investment Banking
arXiv 2609.39958coding agents Published 2026-09-30
Learning When and How to Intervene: A Hindsight-Distilled Sentinel for Coding Agents
arXiv 2609.39957coding agents Published 2026-09-30
ConflictGuide: AutoResearch Improves When Competing Behaviors Are Made Visible
arXiv 2609.39933agent evaluation Published 2026-09-30
OSWorld-Science: A Benchmark of Computer Use Agents for Learning and Using Scientific Software
arXiv 2609.39903quantization Published 2026-09-30
Beyond Accuracy: Prefix-Invariant Realizations of Low-Precision Fast Matrix Multiplication
arXiv 2609.39816quantization Published 2026-09-30
RATIO: Reasoning Analysis and Token-level Inference Optimization for Quantized Reasoning Models
arXiv 2609.39801memory and context Published 2026-09-30
MemCodex: Self-Programming Hierarchical Memory for Language Agents
arXiv 2609.39765efficient training Published 2026-09-30
SE-ADD: Self-Evolving Audio Deepfake Detection with Mistake-Driven Supervision
arXiv 2609.39679safety and guardrails Published 2026-09-30
Aletheia: Permission-Minimality Testing for Coding-Agent Rules
arXiv 2609.39678other Published 2026-09-30
Hybrid Methods for Robust Tabular Data Imputation
arXiv 2609.39613other Published 2026-09-30
GroundingPI: A Grounding Foundation Model towards Physical Intelligence with Visual Primitives
arXiv 2609.39601failure recovery Published 2026-09-30
Thinking Outside the Box: Can Language Models Rely on External Guidance Selectively?
arXiv 2609.39578agent evaluation Published 2026-09-30
A2Z GameSpec-Bench: How Faithfully Can Coding Agents Generate Games from Game Design Specifications?
arXiv 2609.39564multi-agent orchestration Published 2026-09-30
RankEvolve: A Reliable Multi-Agent Auto-Research Harness for Evolving Ranking Models
arXiv 2609.39551safety and guardrails Published 2026-09-30
Speculative Safety Honeypot: Toward Proactive Defense Against Multi-turn Agent Attacks
arXiv 2609.39549efficient training Published 2026-09-30
Comparative study of adapting pre-trained models for driving behavior video captioning
arXiv 2609.39542efficient training Published 2026-09-30
Towards Robust Time Series Learning via Capacity-Centric Modulation
arXiv 2609.39489safety and guardrails Published 2026-09-30
ActionGuard: Tool Call Authorization under Poisoned Skills
arXiv 2609.39450other Published 2026-09-30
Distributionally robust linear regression through the lens of adversarial training
arXiv 2609.39449efficient training Published 2026-09-30
QuantCode Model: Specializing Language Models for Executable Algorithmic Trading Code
arXiv 2609.39420efficient training Published 2026-09-30
Exploring Heterogeneous Model Merging Approach for Complex Knowledge Transfer
arXiv 2609.39369efficient training Published 2026-09-30
A Dynamical Theory of LoRA in Continual Learning
arXiv 2609.39367tool use Published 2026-09-30
Rethinking Multi-Image Re-Representation in Multi-Image Understanding
arXiv 2609.39363attention and kernels Published 2026-09-30
LampAttention: Look-Ahead Mixed-Precision FlashAttention for Dedicated Accelerators
arXiv 2609.39361safety and guardrails Published 2026-09-30
Hiding in Plain Sight: Decoupling Pretext from Actuation for Skill Poisoning in LLM Agents
arXiv 2609.39352efficient training Published 2026-09-30
HAPMoE: Heterogeneity-Aware Automatic Parallelism Planning for Mixture-of-Experts Models Training
arXiv 2609.39350serving and batching Published 2026-09-30
Taming Speculative Search for Test-Time Scaling in LLM Serving
arXiv 2609.39334KV cache Published 2026-09-30
PatchKV: Weight-Space Compensation of KV Cache
arXiv 2609.39329efficient training Published 2026-09-30
WorkGenesis: Building the Worlds That Teach Agents to Work
arXiv 2609.39325distillation Published 2026-09-30
ReSAIL: Mitigating Collapse in Iterative Agent Self-Distillation
arXiv 2609.39306coding agents Published 2026-09-30
Scale and Selection: What Makes Automatic Harness Evolution Work for Visual-Interface Robot Agents
arXiv 2609.39304quantization Published 2026-09-30
QATFactory: A Versatile, Deployment-Aligned Framework for Quantization-aware Training and Distillation of LLMs
arXiv 2609.39223KV cache Published 2026-09-30
Low-Discrepancy Dither for Quantized Recurrent State Caches
arXiv 2609.39185efficient training Published 2026-09-30
ID Balancing: Stable Training of Extremely Sparse MoE via PID-Based Load Control
arXiv 2609.39137KV cache Published 2026-09-30
Characterizing High Bandwidth Flash for LLM Serving
arXiv 2609.39131planning and decomposition Published 2026-09-30
CORE: Conflict-Oriented Reasoning Elimination for Verifiable Language-Model Search
arXiv 2609.39069KV cache Published 2026-09-30
SparseEngine: Sparse-First Inference Engine
arXiv 2609.39068efficient architectures Published 2026-09-30
Switching Linear Attention
arXiv 2609.39034serving and batching Published 2026-09-30
Vosti: Specifying, Implementing, and Verifying Deterministic LLM Inference
arXiv 2609.38981serving and batching Published 2026-09-30
Adaptive Self-Consistency: From Black-Box Sampling to Distribution-Valued Feedback
arXiv 2609.38931other Published 2026-09-30
CellMSA: Context Modeling for Single-Cell Representation Learning
arXiv 2609.38908other Published 2026-09-30
Unmerge: Efficient Machine Unlearning via Task Arithmetic
arXiv 2609.38895sparsity and pruning Published 2026-09-30
Audio Token Attention Is Predictable Before the Language Model Runs
arXiv 2609.38878safety and guardrails Published 2026-09-30
SparLeak: Privacy Leakage from Sparse Attention in LLM Inference on Shared GPUs
arXiv 2609.38830efficient training Published 2026-09-30
DCM-SAM: Defect-Conditioned Mixture of LoRA Experts for NPU-Deployed AM Defect Segmentation
arXiv 2609.38811speculative decoding Published 2026-09-30
Recovering Off-Policy Supervision for Speculative Decoding
arXiv 2609.38795quantization Published 2026-09-29
STEPQuant: When and Where Errors Matter in Delta-Rule Recurrent State Quantization
arXiv 2609.38169quantization Published 2026-09-29
LeapQuant: Efficient Linear Attention with Accurate Recurrent State Quantization
arXiv 2609.38166multi-agent orchestration Published 2026-09-29
Thinking Before Thinking: Scaling Agentic Inference Through Meta-Reasoning
arXiv 2609.38147multi-agent orchestration Published 2026-09-29
Learning Meta-Skills for Agent Harness Design in Test-Time AI4AI
arXiv 2609.38143KV cache Published 2026-09-29
WUSH-KV: KV Cache Quantization with Data-Adaptive Transforms
arXiv 2609.38121sparsity and pruning Published 2026-09-29
Pruning for Efficiency, Paying in Fairness: Demographic Disparities in Pruned Speech-LLMs
arXiv 2609.38106serving and batching Published 2026-09-29
Mira: Memory-Efficient MoE Inference Using Adaptive Caching and Predictive Expert Staging
arXiv 2609.38090agent evaluation Published 2026-09-29
UserProxyBench: Evaluating LLM User Simulators for Agent Benchmarks and Training
arXiv 2609.38043tool use Published 2026-09-29
Retrieval-Augmented Skill Optimization via Cross-Harness Adaptation
arXiv 2609.38024coding agents Published 2026-09-29
Merged, Not Measured: An Empirical Study of Performance Issues Fixed by Coding Agents
arXiv 2609.37985coding agents Published 2026-09-29
SelfSearch: Reward-Free Search for Self-Improving Agents
arXiv 2609.37968serving and batching Published 2026-09-29
RLX: A Unified Multi-Backend Tensor Compiler and Distributed Runtime in Rust
arXiv 2609.37916quantization Published 2026-09-29
Behavioral Capacity Certificates for Quantized Language Models
arXiv 2609.37887agent evaluation Published 2026-09-29
AgentBug-Smith: Automatically Reproducing Real-World Harness Bugs in Agentic Systems
arXiv 2609.37864safety and guardrails Published 2026-09-29
Storage Is Not Strategy: State-Conditioned Support Control for LLM Unlearning
arXiv 2609.37858efficient training Published 2026-09-29
Scaling Influence Functions in LLMs through Eigenbasis-Corrected One-Bit Gradient Projection
arXiv 2609.37842serving and batching Published 2026-09-29
Joint Effects of GPU Server Topology, Parallelism, and Congestion Control on MoE Inference: A Controlled Simulation Study
arXiv 2609.37828efficient architectures Published 2026-09-29
Thinking in Depth, Speaking Directly: Recurrent Latent Reasoning for Paralinguistically Grounded Spoken Dialogue
arXiv 2609.37818memory and context Published 2026-09-29
ContextRender: From Execution Dependencies to Agent Context
arXiv 2609.37743other Published 2026-09-29
Foundation Neural-Network Quantum States for Molecular Potential Energy Surfaces in Second Quantization
arXiv 2609.37733memory and context Published 2026-09-29
Context Language Models
arXiv 2609.37725sparsity and pruning Published 2026-09-29
Predictive Geometry of Hidden Trajectories in Transformers
arXiv 2609.37717other Published 2026-09-29
Zephyr: An Efficient Audio Denoising System Using Spiking Neural Networks Enabled With A Sparsity-Aware Flexible FPGA PE Array
arXiv 2609.37711agent evaluation Published 2026-09-29
EngiWorld: What Can Frontier Agents Deliver in Professional Engineering Environments?
arXiv 2609.37686retrieval and RAG Published 2026-09-29
KUPAS MASTER: Distilling the Tacit Expertise of Master Practitioners into Agent-Ready Experience Corpora
arXiv 2609.37673agent evaluation Published 2026-09-29
EnterpriseBench: Benchmarking LLM Agents on Enterprise-Level Strategic Reasoning and Decision-Making
arXiv 2609.37658efficient training Published 2026-09-29
RLTL;DR: Self-improvement by Internalizing Self-generated Feedback
arXiv 2609.37633serving and batching Published 2026-09-29
SPLASH: Switching Parallel Layouts of Attention with Seamless Handoff for LLM Serving
arXiv 2609.37626other Published 2026-09-29
GraphVQ: Structure-Aware Autoregressive Decoding over Context-Quantized Graph Tokens
arXiv 2609.37604memory and context Published 2026-09-29
FOCUS: Training-Free Decision-Preserving Context Compression for LLM Agents
arXiv 2609.37590efficient training Published 2026-09-29
SkillGym: Training Skill-Use Agents with Automatic Verifiable Environment Generation
arXiv 2609.37539speculative decoding Published 2026-09-29
DScale: Scaling Block-Diffusion Speculative Decoding with Adaptive Verification
arXiv 2609.37532agent evaluation Published 2026-09-29
Hierarchical Compression of Vision-Language Model Benchmarks
arXiv 2609.37515quantization Published 2026-09-29
Scale Sensitivity in Low-Bit Post-Training Quantization: Curvature of the Quantization Error Landscape
arXiv 2609.37416other Published 2026-09-29
Encore: Few-Shot Agentic Discovery of Manipulation Strategies
arXiv 2609.37359agent evaluation Published 2026-09-29
Do Agent Benchmarks Do What They Say? An Executable-Contract Audit of Tool-Using Agent Environments
arXiv 2609.37315agent evaluation Published 2026-09-29
Foundations of Proactive Agents: Principles, Technical Layers, and Proactivity-Gym
arXiv 2609.37267agent evaluation Published 2026-09-29
DatalogBench: Evaluating Large Language Models on Text-to-Datalog Synthesis
arXiv 2609.37233retrieval and RAG Published 2026-09-29
Follow the Entities: A Corpus Map for Agentic Search
arXiv 2609.37226safety and guardrails Published 2026-09-29
ToolFence: Fine-Grained Authorization for Secure Tool-Using LLM Agents
arXiv 2609.37196tool use Published 2026-09-28
TokenCast: Forecasting Token Consumption During LLM Agent Execution
arXiv 2609.35760KV cache Published 2026-09-28
KV-streams for Efficient Compaction in Agentic Reinforcement Learning
arXiv 2609.35750tool use Published 2026-09-28
Harness Learning Enables Generalizable Test-Time Adaptation
arXiv 2609.35738tool use Published 2026-09-28
Failure-Transparent Agents: Benchmarking Post-Failure Reporting in Tool-Using Language Models
arXiv 2609.35732agent evaluation Published 2026-09-28
GPUPhysBench: Benchmarking Coding Agents for Correct and Efficient GPU Physics Simulation
arXiv 2609.35639attention and kernels Published 2026-09-28
SANTA++: Sampling Attention through Representative Keys
arXiv 2609.35629other Published 2026-09-28
Arbitrary-Accuracy Neural Approximation with Optimal Neuron Count and Near-Optimal Bit Complexity
arXiv 2609.35628KV cache Published 2026-09-28
Cartridges++: KV Cache Compression without Off-Context Derailment
arXiv 2609.35621efficient training Published 2026-09-28
EvE: An Alternate Optimizer to Adam
arXiv 2609.35614other Published 2026-09-28
Learning Conditional Expectation Operators via Functional Newton Updates
arXiv 2609.35598safety and guardrails Published 2026-09-28
SEABench: Benchmarking Endogenous Misalignment In Self-Evolving Agents
arXiv 2609.35596other Published 2026-09-28
Language Models Act on Hidden Valence
arXiv 2609.35591sparsity and pruning Published 2026-09-28
Output-aware Residual Stream Pruning for Large Language Models
arXiv 2609.35579memory and context Published 2026-09-28
FactorEngram: Factorized N-gram Memory with Basis-Level Gating for Language Models
arXiv 2609.35578planning and decomposition Published 2026-09-28
From Search to Research: Exploring Search Scaling in Autonomous Quantitative Factor Mining
arXiv 2609.35559safety and guardrails Published 2026-09-28
The Compiler May Read It, the Agent May Not: Keeping Part of a Research Code Away from a Coding Agent
arXiv 2609.35557memory and context Published 2026-09-28
Continuous Context Management
arXiv 2609.35540efficient training Published 2026-09-28
ARISE: Adapting to Evolving Capability Gaps in Agentic Reinforcement Learning
arXiv 2609.35532coding agents Published 2026-09-28
AutoRef: Harness Optimization for Agentic Multi-Reference Image Generation
arXiv 2609.35530other Published 2026-09-28
Structured Latent Modeling for Supervised Multimodal Information Decomposition
arXiv 2609.35502quantization Published 2026-09-28
Tetra: Serving Leech-Lattice Quantized LLMs at 2.7 Bits per Parameter
arXiv 2609.35465tool use Published 2026-09-28
A.D.A.M.O. (Agent for language-Driven Actions with Multimodal Observations): A Visual-Symbolic Framework for Virtual Humans
arXiv 2609.35463memory and context Published 2026-09-28
LLMs are General Asynchronous Agents
arXiv 2609.35427coding agents Published 2026-09-28
Semantic Prefix Oracles for LLM Decoding: Contracts and Differential Validation
arXiv 2609.35425failure recovery Published 2026-09-28
MCP Error Messages Written for Developers Hurt the Most Capable Agents Most
arXiv 2609.35381other Published 2026-09-28
From Pixel to Poses: Object-centric Tool Manipulation Learning from Human Demonstrations
arXiv 2609.35375other Published 2026-09-28
From Input to Output: A Flexible Agent for Dual-End Interpretation of Sparse Autoencoder Features
arXiv 2609.35367failure recovery Published 2026-09-28
Planarian: Managing Agent State with Statepoints
arXiv 2609.35366agent evaluation Published 2026-09-28
Do Coding Agents Reuse Existing Code or Reinvent the Wheel?
arXiv 2609.35357quantization Published 2026-09-28
Fiona: Accelerating FHE Inference with Packing-Aware Ternary Weights
arXiv 2609.35352multi-agent orchestration Published 2026-09-28
TMCS: Tool-Grounded Multi-Agent Reasoning for Compositional Chemical Problem Solving
arXiv 2609.35336coding agents Published 2026-09-28
Hyper Algorithm Design Agent: Evolving Learnable Optimizer from Zero
arXiv 2609.35328agent evaluation Published 2026-09-28
Reliability Engineering for AI Systems: Challenges, Methods, and Directions
arXiv 2609.35316other Published 2026-09-28
Latency and accuracy tradeoffs in Spiking Neural Networks
arXiv 2609.35260distillation Published 2026-09-28
On-Policy or Off-Policy Learning? A Systematic Study of Distillation Dynamics
arXiv 2609.35259efficient architectures Published 2026-09-28
Beyond Selection: Token Parameterization for Extreme Visual Token Compression
arXiv 2609.35232speculative decoding Published 2026-09-28
Beneath the Tokens: A Performance Engineering Study of Multi-Token Prediction in GPU-Accelerated LLM Inference
arXiv 2609.35188KV cache Published 2026-09-28
CacheRepair: Learning to Repair Cross-Chunk Context in RAG for KV Cache Fusion
arXiv 2609.35139attention and kernels Published 2026-09-28
E3J: An Efficient and Open-Source Backend for Euclidean Equivariant Operations on GPU and TPU
arXiv 2609.35099other Published 2026-09-28
Continuous Variational Synthesis
arXiv 2609.35083coding agents Published 2026-09-26
Beyond the Model: Demystifying Harness Effects in Software Engineering Agents
arXiv 2609.32459efficient training Published 2026-09-25
New LoRA Skills Should Read but Never Write
arXiv 2609.31600agent evaluation Published 2026-09-25
AgentWorld: Benchmarking Long-Horizon Collaboration of Multi-agent LLMs
arXiv 2609.31590coding agents Published 2026-09-25
Compact Documentation for Coding Agents: A Benchmark, an Optimizer, and Why It Does Not Transfer
arXiv 2609.31587memory and context Published 2026-09-25
DeepEdu-v1: Efficient and Scalable Agentic LLMs for Vietnamese Education
arXiv 2609.31568quantization Published 2026-09-25
Generalization behavior of OPTQ and the role of regularization
arXiv 2609.31560serving and batching Published 2026-09-25
EAServe: Encode-Aware Disaggregated Serving for Multimodal Large Language Models
arXiv 2609.31551efficient training Published 2026-09-25
ViSTA: A Simple Bridge Extends Visual Alignment to Clinical Time-Series Understanding in Multimodal LLMs
arXiv 2609.31448multi-agent orchestration Published 2026-09-25
Towards Mitigating Fabricated Consensus: The Active Provenance Gate for Multi-Agent Debate Synthesis
arXiv 2609.31422KV cache Published 2026-09-25
Evaluating the accuracy of KV cache reuse techniques
arXiv 2609.31415retrieval and RAG Published 2026-09-25
Intent2Tc: Automated Intent-to-Traffic Control Translation with Language Models
arXiv 2609.31397quantization Published 2026-09-25
Towards Understanding LLM-Based Log Anomaly Detection: An Empirical Study of Performance, Efficiency, and Robustness
arXiv 2609.31371safety and guardrails Published 2026-09-25
A Safety-Bounded SDC-to-MCP Gateway for Medical AI Agents
arXiv 2609.31358serving and batching Published 2026-09-25
The Right Information Extraction Pipeline Depends on the Document: Accuracy-Energy Trade-offs for Small, Local Models
arXiv 2609.31341safety and guardrails Published 2026-09-25
AgentXploit: Autonomous Repository-to-Runtime Red-Teaming for AI Agents
arXiv 2609.31318other Published 2026-09-25
LUCID: Learning Under Confounding for Inference and Discovery in Time Series
arXiv 2609.31315quantization Published 2026-09-25
Softmax Reparameterization for Output-Head Quantization
arXiv 2609.31291efficient architectures Published 2026-09-25
MoSAR: Mixture of Semantic Attention Regimes for Learning Adaptive and Approximable Attention Geometries
arXiv 2609.31261memory and context Published 2026-09-25
PIA: A Personal Intelligence Agent Turning Health Conversations into Records and Records into Understanding
arXiv 2609.31255coding agents Published 2026-09-25
Semantic Navigation for Issue Localization in Code Repository
arXiv 2609.31176efficient training Published 2026-09-25
Audio emotion recognition for atypical hearing
arXiv 2609.31168other Published 2026-09-25
AgentRecommender: LLM Agents Enable Customizable Recommender Systems on the User Side
arXiv 2609.31166distillation Published 2026-09-25
Momentum-Guided Federated Split Distillation for Personalized Temporal Edge Intelligence
arXiv 2609.31159other Published 2026-09-25
Bayesian Tensor Autoencoder with Physics-informed Predictive Prior for Multi-dimensional Time Series Anomaly Detection
arXiv 2609.31157other Published 2026-09-25
The Residual Stream's Effective Depth
arXiv 2609.31098safety and guardrails Published 2026-09-25
CG-Probes: Recovering Guardrail Directions from Patient Query Embeddings
arXiv 2609.31062efficient training Published 2026-09-25
Distributed Learning as a Service: The Developer's Perspective
arXiv 2609.31061serving and batching Published 2026-09-25
DynBranch: Speculative Subgraph Reuse for Dynamic Agentic LLM Serving
arXiv 2609.31047safety and guardrails Published 2026-09-25
MetaPermit: Scalable and Auditable Access Control for AI Agents via LLM-Inferred Meta-Attributes
arXiv 2609.31039quantization Published 2026-09-25
G$^2$PTQ: Improving LLM Post-Training Quantization with Generalized Gradient Compensation
arXiv 2609.31009quantization Published 2026-09-25
Low-Bit Recurrent States in Hybrid Language Models
arXiv 2609.30950multi-agent orchestration Published 2026-09-25
Financial Fragility in Societies of LLM Agents: Coordination Failures and Stabilizing Mechanisms
arXiv 2609.30940tool use Published 2026-09-25
ToolSearcher: Optimizing Tool Selection at Scale via Reinforcement Learning
arXiv 2609.30906other Published 2026-09-25
Retraction-Based Gradient Projection Algorithms on Manifolds
arXiv 2609.30885KV cache Published 2026-09-25
CacheReforge: Bounded Recovery for Stale KV Caches under Evolving Adapters
arXiv 2609.30884safety and guardrails Published 2026-09-25
AGATE: Provenance-Based Runtime Defense Against Compositional Attacks on LLM Agents
arXiv 2609.30830agent evaluation Published 2026-09-25
A Benchmark and Diagnostic Study of Epistemic Admission in Shared Agent Memory
arXiv 2609.30813agent evaluation Published 2026-09-25
Evaluating Real-Time Voice Agents: From Component Quality to Grounded Outcomes
arXiv 2609.30798coding agents Published 2026-09-25
Analyzing and Mitigating Cost-Inefficient Behaviors in Coding Agents
arXiv 2609.30725safety and guardrails Published 2026-09-25
Prompt Injection Detection for Email Agents Through Attack Chain Modeling
arXiv 2609.30657safety and guardrails Published 2026-09-24
Subjects, Not Authors: The Authorship Hazard in Agentic Dataspaces
arXiv 2609.30614safety and guardrails Published 2026-09-24
LLM Agents Can Easily Tamper With Their Own Traces
arXiv 2609.30266coding agents Published 2026-09-24
RAPID: Robot Agentic Programming from Demonstrations
arXiv 2609.30249coding agents Published 2026-09-24
Coding Agents for Generalized Task and Motion Planning Problems
arXiv 2609.30233efficient training Published 2026-09-24
PoEM: Predicting RL Outcomes from Existing Policies
arXiv 2609.30226safety and guardrails Published 2026-09-24
Instrumental Monitor Evasion Emerges Under Ordinary Task Pressure
arXiv 2609.30217agent evaluation Published 2026-09-24
ExplorationBench: Measuring AI Systems' Exploration in Verifiable Alien Worlds
arXiv 2609.30199agent evaluation Published 2026-09-24
Evaluating Agent Skills for Version-Specific Plugin Migration: A Retrospective Study
arXiv 2609.30120other Published 2026-09-24
AT-SKM-Net: An Accelerated Trainable Sampling Kaczmarz-Motzkin Framework for Linear Hard-Constraint Feasibility on Dynamic Graphs
arXiv 2609.30088attention and kernels Published 2026-09-24
KernelOPT: Dispatch-Aware Agentic Search for GPU Kernel Optimization
arXiv 2609.30059quantization Published 2026-09-24
AERIAL: Adversarial Evaluation of Robustness in Accuracy-Preserving Low-Precision EEG Decoders
arXiv 2609.30037safety and guardrails Published 2026-09-24
How does Adversarial Influence Scale in Multi-Agent Systems?
arXiv 2609.30028agent evaluation Published 2026-09-24
Synthetic Hospital: An Open, Verifiable, Physician-Validated Longitudinal EHR Benchmark
arXiv 2609.30027agent evaluation Published 2026-09-24
Low-Cost Assays for Measuring Model Behavior Across Vendors and Releases
arXiv 2609.30012retrieval and RAG Published 2026-09-24
Automated Regulatory Compliance Question Answering in Financial Services with Domain-Adapted Retrieval-Augmented Generation
arXiv 2609.30009quantization Published 2026-09-24
GHOST-Q: Towards Studying Grounding Hallucinations Overlooked Under Same-score TradeOffs in Quantized VLMS
arXiv 2609.29999safety and guardrails Published 2026-09-24
Guardrails or Roadblocks? Effects of Pedagogical Style and Context Awareness in AI Teaching Assistants for Programming
arXiv 2609.29995multi-agent orchestration Published 2026-09-24
World Action Agent: Harnessing VLMs for Robot Manipulation via World Action Rehearsal
arXiv 2609.29964KV cache Published 2026-09-24
MILO: Efficient Many-shot In-Context Learning with Block-wise Low-rank Compression
arXiv 2609.29913other Published 2026-09-24
Spatio-temporally complementary feature propagation on graphs for longitudinal AADT estimation
arXiv 2609.29906efficient training Published 2026-09-24
Qwen-Planner-Agent: A Closed-Loop AI-for-AI Framework for Real-World Mobile Planner Agents
arXiv 2609.29892memory and context Published 2026-09-24
When Can Agents Forget Their Reasoning? ICLR for Long-Horizon Agent Context Compression
arXiv 2609.29875efficient architectures Published 2026-09-24
Does per-frame early exit pay? A compute-matched study of dynamic depth for on-device speech enhancement
arXiv 2609.29867efficient architectures Published 2026-09-24
Beyond Model Size: Redesigning LiSenNet for embedded speech enhancement
arXiv 2609.29866tool use Published 2026-09-24
PUBG Ally: A Conversational Embodied Agent as an AI Teammate
arXiv 2609.29837KV cache Published 2026-09-24
FlashLoop: Fast and Memory-Efficient Looped Transformers via Lazy Updates
arXiv 2609.29812safety and guardrails Published 2026-09-24
Hard Stop: Kernel-Level Preemption and Containment for Rogue Agentic Execution
arXiv 2609.29808efficient training Published 2026-09-24
Adaptive Fisher-Whitened Cross-Covariance for Low-Resource Speech Recognition
arXiv 2609.29800agent evaluation Published 2026-09-24
Breaking the Environment Wall: Evolving LLM Agent Environments for Recursive Self-Improvement
arXiv 2609.29773attention and kernels Published 2026-09-24
Dense Matrices Are Alike; Sparse Matrices Are Sparse in Their Own Way: A Structure-Adaptive Tile Cholesky Factorization
arXiv 2609.29765other Published 2026-09-24
On Growth and Form, and Function: Reusable Regulatory Handles Control Phenotypic Variation
arXiv 2609.29755coding agents Published 2026-09-24
Between the Commits: Process, Error, and Claim Reliability in a Wholly AI-Authored Codebase
arXiv 2609.29744other Published 2026-09-24
VQ-LIC: Shared Vector-Quantized Learned Image Compression on a Resource-Constrained FPGA
arXiv 2609.29727agent evaluation Published 2026-09-24
SWE-Prometheus: Measuring Engineering Governance Improvements in Real-World Repositories
arXiv 2609.29465tool use Published 2026-09-24
Demystifying Agent Skills for Smart Contract Auditing: Design, Effectiveness, Behavioral Impact
arXiv 2609.29454planning and decomposition Published 2026-09-24
IterSynth: Rethinking Deep Search Agents via Role-Decoupled Iterative Synthesis
arXiv 2609.29444efficient training Published 2026-09-24
Rufus-Air: An Open LLM Post-Training Recipe
arXiv 2609.29421sparsity and pruning Published 2026-09-24
Resource-Aware Model Selection for Scalable Indoor Localization on HPC Platforms
arXiv 2609.29402quantization Published 2026-09-24
Baseline Shape Decides the Verdict: A Controlled Re-Examination of Ternary Language Models at 60K Parameters
arXiv 2609.29397distillation Published 2026-09-24
Decoupled Early Exits for Task-Dependent Compute Allocation in Flow-Matching VLAs
arXiv 2609.29382other Published 2026-09-24
BanglaTurn: A Benchmark and Whisper-Based Model for End-of-Turn Detection in Bangla Speech
arXiv 2609.29371safety and guardrails Published 2026-09-24
SkinAgent AI: A Safety-Grounded Multimodal Agentic Framework for Non-Diagnostic Skincare Support
arXiv 2609.29341tool use Published 2026-09-24
Policy as Code: A Coroutine-Bridge Harness for Fast-Reasoning Reliability on CAR-bench
arXiv 2609.29251distillation Published 2026-09-24
Continuous Online Fault Detection for Mobile Robots via Adaptive Edge Models
arXiv 2609.29194agent evaluation Published 2026-09-24
IndicBankBench: Evaluating Safety and Reliability of Language Model Assistants in Indian Retail Banking
arXiv 2609.29167serving and batching Published 2026-09-24
Cross-Model Autoscaling for Shared LLM Serving
arXiv 2609.29160failure recovery Published 2026-09-24
A Wrong Turn Does Not Ruin the Journey: Deviation-Guided Skill Self-Evolution for LLM Agents
arXiv 2609.29154memory and context Published 2026-09-24
Scope Before You Persist: Preventing Cross-Family Interference in Agent Memory
arXiv 2609.29144failure recovery Published 2026-09-24
Where Does Exactly-Once Live? Model, Harness, and Tool-Contract Effects on Duplicate Side Effects in LLM Agents
arXiv 2609.29095other Published 2026-09-24
A Rapid Pipeline for Training and Deploying ML Models on WeBe Band
arXiv 2609.29084efficient training Published 2026-09-24
From Self-Distillation to Self-Practice: Privileged Information for Multi-Turn Agents
arXiv 2609.29051tool use Published 2026-09-24
SLCA-GRPO: Resolving Cross-Segment Credit Misattribution in Tool-Calling RL
arXiv 2609.29050safety and guardrails Published 2026-09-24
Where Hallucinations Live: A Cross-Architecture Circuit in VQ-Tokenized Vision-Language Models
arXiv 2609.29048multi-agent orchestration Published 2026-09-24
Multi-Agent Orchestration of 3GPP Channel Estimators
arXiv 2609.29044sparsity and pruning Published 2026-09-24
Paging the Experts: A Reproducible Characterization of Flash-Backed MoE Inference on iPhone
arXiv 2609.29032sparsity and pruning Published 2026-09-24
Exploiting answer-invariant redundancies in satellite imagery for efficient VLM inference on edge
arXiv 2609.29029agent evaluation Published 2026-09-24
When Does Action Credit Need Updating?
arXiv 2609.29007efficient training Published 2026-09-24
Automatic Rank Allocation for Low-Rank Adaptation in Large Language Models via lp Regularization
arXiv 2609.28998agent evaluation Published 2026-09-24
Beneath the Scores: Rethinking Hallucination Evaluation for Video Understanding Models
arXiv 2609.28991efficient training Published 2026-09-24
Personalized Korean Lipreading as Visual Speech Recognition: Transfer, Census and Adaptation on OLKAVS
arXiv 2609.28988quantization Published 2026-09-24
Same Bit Width, Different Outcomes: Post-Training Quantization of Text-to-Speech Across Architectures
arXiv 2609.28974efficient training Published 2026-09-24
Back to the Definition: Estimating Step-Level Advantages via Trajectory Graphs for Agentic Reinforcement Learning
arXiv 2609.28963coding agents Published 2026-09-24
Control the Harness, Control the Cost: Routing and Governing AI Coding Agents in the Enterprise
arXiv 2609.28919safety and guardrails Published 2026-09-24
On the Effectiveness of Kernel-Level Evidence for Agent Security
arXiv 2609.28915efficient architectures Published 2026-09-24
Multimodal Routing and Region Refinement for Language-Guided Medical Image Segmentation
arXiv 2609.28860other Published 2026-09-23
Physics-Guided Multi-Objective Deep Learning for Ultrasound RF Data Interpolation in Resource-Constrained Imaging
arXiv 2609.28775distillation Published 2026-09-23
Small yet Assistive: Spatially-Aware Post-Training for Low Vision
arXiv 2609.28757coding agents Published 2026-09-23
Soundness Checking of Taint Flow Models
arXiv 2609.28750sparsity and pruning Published 2026-09-23
Temporal Taxation Compounds Under Post-Training Compression of Whisper Models
arXiv 2609.28739safety and guardrails Published 2026-09-23
Spooftral: Can Voxtral Audio-Language Model Detect Speech Spoofing?
arXiv 2609.28713safety and guardrails Published 2026-09-23
Progressive Skill Discovery as Access Control for Tool-Using LLM Agents: Structural Governance through Role-Scoped Capability Delivery
arXiv 2609.28693safety and guardrails Published 2026-09-23
Agent Approval Laundering: Transitive Effects Beyond the Approved Invocation
arXiv 2609.28586safety and guardrails Published 2026-09-23
Persistent Billable State: Denial-of-Wallet Attacks and Defenses in Tool-Calling LLM Agents
arXiv 2609.28585other Published 2026-09-23
SpaFactor: Lightweight Spatial Context-Aware Gene Program Modeling for Histology-to-Transcriptomics Inference
arXiv 2609.28563safety and guardrails Published 2026-09-23
Who Is Behind the Harness? Fingerprinting LLMs through Agentic Behavior
arXiv 2609.28559other Published 2026-09-23
CFD Correction of Open Tip Clearance Flow in a Compressor Cascade Using VAE Latent Space Adaptation
arXiv 2609.28558planning and decomposition Published 2026-09-23
Pistis Technical Report
arXiv 2609.28554other Published 2026-09-23
Minimal-Norm Univariate Two-Layer ReLU Classification: Exact Solutions and Global Optimality with Skip Connections
arXiv 2609.28438efficient training Published 2026-09-23
Cross-Scale Transfer Learning for Depression Severity Prediction: From PHQ-8 to HAMD-17 Across Languages and Clinical Paradigms
arXiv 2609.28430agent evaluation Published 2026-09-23
Shopping by algorithm: How agentic AI deploys human heuristics as a surrogate consumer
arXiv 2609.28372quantization Published 2026-09-23
MicroQonv: Reshaping Convolution Tensors for Efficient Microscaling in Training and Inference
arXiv 2609.28358efficient architectures Published 2026-09-23
Non-Commutative State Tracking with Input-Dependent Low-Rank Updates in Mamba-3
arXiv 2609.28273quantization Published 2026-09-23
Predicting Quantization Price for Selecting PTQ Configurations Before Deployment
arXiv 2609.28270quantization Published 2026-09-23
RAMP: Robust Adaptive Mixed-Precision Quantization for Edge CPU Vision Models
arXiv 2609.28262memory and context Published 2026-09-23
Complementary Roles of Activation and Parametric Memory in Few-Shot Learning
arXiv 2609.28250other Published 2026-09-23
hyperbolix: Hyperbolic Deep Learning in JAX
arXiv 2609.28248other Published 2026-09-23
Confidence Falls Short: Asymmetric Certainty Gains from Optimization Hinder Multimodal Classification
arXiv 2609.28165KV cache Published 2026-09-23
Tensor Decomposition of Transformer Key-Value Caches: Spectral Structure and Format Comparison
arXiv 2609.28029sparsity and pruning Published 2026-09-23
Task-Induced Riemannian Metrics for Vision Transformer Feature Spaces
arXiv 2609.27988KV cache Published 2026-09-23
Risk-Controlled KV-Cache Eviction: From Memory Budgets to Risk Targets
arXiv 2609.27981sparsity and pruning Published 2026-09-23
Six Layers Less: Encoder Pruning for Whisper with Label-Free Recovery
arXiv 2609.27980KV cache Published 2026-09-23
The KV Cache Working Set: Online Capacity Planning for LLM Inference Systems
arXiv 2609.27746efficient architectures Published 2026-09-23
NS-ATTENTION: Newton-Schulz Transformations of Attention Outputs in Vision Transformers
arXiv 2609.27735distillation Published 2026-09-23
SkillGym: Internalizing Human Skills into LLMs for Real-World Problem Solving
arXiv 2609.27717other Published 2026-09-23
What Do Tabular Foundation Models Compute In Context? In-Situ Representation Refinement through Attention-Gated Updates
arXiv 2609.27679safety and guardrails Published 2026-09-23
InGuard: Towards Generalized Inner Guardrail for Safe Text-to-Image Generation
arXiv 2609.27620memory and context Published 2026-09-23
State-Grounded Conditioning: Wrapping User-Facing LLM Agents Where Direction Depends on Live State
arXiv 2609.27606other Published 2026-09-23
The Capability Manifold and ML Scaling Laws
arXiv 2609.27588coding agents Published 2026-09-23
Unity Insight: A Production Code--Asset Index for LLM Coding Agents in Unity Projects
arXiv 2609.27585efficient training Published 2026-09-23
ProCredit: From Outcome Rewards to Progress Credit in Agentic Reinforcement Learning
arXiv 2609.27532KV cache Published 2026-09-23
DeltaS: Reading the Gated Linear Attention State for KV Cache Eviction in Streaming Video
arXiv 2609.27470agent evaluation Published 2026-09-23
What Looks Like a Capability Limit in Vision-Language Models Is a Readout Limit
arXiv 2609.27408speculative decoding Published 2026-09-23
When Parallel Drafter Meets Parallel Speculative Decoding
arXiv 2609.27396safety and guardrails Published 2026-09-23
Psychoacoustically Aligned Latent Smoothing for Adversarial Robustness of Full-Duplex Speech-to-Speech Dialogue Models
arXiv 2609.27378quantization Published 2026-09-23
Quantization-Robust Unlearning through the Lens of Retain-Forget Loss Landscapes Interaction
arXiv 2609.27355memory and context Published 2026-09-23
Constraint-Driven Context Engineering: Designing Domain Interfaces for AI Systems
arXiv 2609.27354agent evaluation Published 2026-09-23
MolDesignBench: Evaluating LLM-based Agent for Scenario-grounded Molecular Design
arXiv 2609.27349safety and guardrails Published 2026-09-23
CART: Closed-Loop Adaptive Red Teaming for Large Language Models
arXiv 2609.27336memory and context Published 2026-09-23
Just-in-Time Memory: Learning to Curate Task-Adaptive Memory for LLM Agents
arXiv 2609.27334agent evaluation Published 2026-09-23
Verifiable Hidden Dynamics Play: Generating Agentic RL Environments from Solved Mechanisms
arXiv 2609.27321distillation Published 2026-09-23
Learn How to Act from Your Own Interactions: On-Policy Self-Distillation for GUI Agents
arXiv 2609.27307memory and context Published 2026-09-23
StateComp: Learning When to Compress History in Long Horizon Agents
arXiv 2609.27298memory and context Published 2026-09-23
Memory Control Signals Emerge Before Action in Long Horizon Agents
arXiv 2609.27286memory and context Published 2026-09-23
EnSIMem: Entity-Structured Indexing for Long-Term Agent Memory
arXiv 2609.27279other Published 2026-09-23
Graph Learning with Spectral Connectivity Priors for Scarce Data
arXiv 2609.27278agent evaluation Published 2026-09-23
CAVEAT: Towards Robust Computer-Use Agents in Incentive-Misaligned Environments
arXiv 2609.27273efficient training Published 2026-09-23
Can One Adapted Model Do It All? Fine-Tuning Strategy Selection for Customer Support LLMs
arXiv 2609.27262tool use Published 2026-09-23
From PyTorch to the NPU: LLM-Agent-Driven Model Conversion Across Heterogeneous Inference Runtimes
arXiv 2609.27249KV cache Published 2026-09-23
Distilling Sequential Computation in Transformer Language Models
arXiv 2609.27233efficient training Published 2026-09-23
Reliable Federated TinyML Deployment for IoT Security
arXiv 2609.27202tool use Published 2026-09-22
TomasuLLM: Out-of-Order Speculative Execution for LLM Agents
arXiv 2609.38201other Published 2026-09-22
Certified Task-Conditioned Active Observability
arXiv 2609.28520multi-agent orchestration Published 2026-09-22
PEARL: A Lightweight Prompt-based Feature Interpreter Framework for Real-Time, Anonymous, and Heterogeneous Collaborative Perception
arXiv 2609.27123memory and context Published 2026-09-22
ChipMEM: Verification-Grounded Memory for EDA Agents
arXiv 2609.27067efficient training Published 2026-09-22
Reinforcement Learning with Decomposed Subtasks
arXiv 2609.27035quantization Published 2026-09-22
GeoRVQ: Decoder-aware geometry for residual-token prediction in physiological signals
arXiv 2609.27018efficient training Published 2026-09-22
CRISP: Scalable Importance-Stratified Coresets for Imbalanced Tabular Learning
arXiv 2609.26962efficient architectures Published 2026-09-22
The Computational Value of Sensory-Aligned Receptive Fields Depends on Neuronal Expressivity
arXiv 2609.26940failure recovery Published 2026-09-22
TwinCheck: Evidence-Grounded Negative-Twin Verification for Stateful Tool Agents
arXiv 2609.26911safety and guardrails Published 2026-09-22
Ajar: Measuring Open Privilege in Agent Defenses
arXiv 2609.26900tool use Published 2026-09-22
Harness as a Language: A Minimalist Agent Framework With Maximal Expressivity
arXiv 2609.26891safety and guardrails Published 2026-09-22
Safety Nudges: User-Facing Interventions for Real-Time AI Risk Awareness
arXiv 2609.26865coding agents Published 2026-09-22
Who Finishes the Job? A Study of Follow-Up Fixes and Commit Authorship on AI Coding Agent Pull Requests
arXiv 2609.26847KV cache Published 2026-09-22
Flash-dLLM: IO-Aware KV Caching and Parallel Decoding for Fast, Memory-Efficient Diffusion LLMs
arXiv 2609.26796other Published 2026-09-22
A Decentralized Partially Observable Team Decision Methodology with Delayed Information Sharing
arXiv 2609.26783multi-agent orchestration Published 2026-09-22
Agensh: Scaling Organizational Intelligence to 1,024 Agents
arXiv 2609.26781agent evaluation Published 2026-09-22
SWE-Serve: Benchmarking Agentic Engineering For Production Inference Serving
arXiv 2609.26777serving and batching Published 2026-09-22
SARA: SLO-Aware Resource Allocation for Disaggregated Agentic LLM Services
arXiv 2609.26763memory and context Published 2026-09-22
Grow the Harness, Not the Context: From Strategy-Free Scaffolds to Reusable Specialist Agents
arXiv 2609.26760quantization Published 2026-09-22
Train Where the Quantized Model Goes: On-Policy Distillation for Low-Bit Reasoning
arXiv 2609.26708agent evaluation Published 2026-09-22
Measuring the Serving Stack Instead of the Model: Hidden Confounds in Local Tool-Use Evaluation
arXiv 2609.26693other Published 2026-09-22
Dynamic Slack-Aware Clocking for Near-Threshold Tensor Processing Units (TPUs)
arXiv 2609.26644other Published 2026-09-22
A Configurable Heuristic for the MLCS Problem
arXiv 2609.26602planning and decomposition Published 2026-09-22
REFLEX with Jev for Efficient Selective Control in LLM Agents
arXiv 2609.26532safety and guardrails Published 2026-09-22
From Approval to Execution: Reconstruction-Aware Repair Analysis for LLM-Agent Software
arXiv 2609.26529other Published 2026-09-22
PP-Net: A Hybrid Physical-Prior Neural Network for Scattered Light Removal in Biomedical Images on Embedded Devices
arXiv 2609.26474planning and decomposition Published 2026-09-22
Combining Hierarchical Cognitive Process with Process Supervision for Interpretable Scene Safety Understanding
arXiv 2609.26399KV cache Published 2026-09-22
HySparse2: Hybrid Sparse Attention with Two-Level KV Sharing
arXiv 2609.26368efficient training Published 2026-09-22
PACT: From Credit Assignment to Critic Alignment
arXiv 2609.26355quantization Published 2026-09-22
Geometry-Aware Hyperbolic Residual Quantization
arXiv 2609.26342quantization Published 2026-09-22
Disaggregated Quantization: Specializing LLM Prefill and Decode
arXiv 2609.26333safety and guardrails Published 2026-09-22
Design and Evaluation of a Controlled Post-Alert Incident Orchestration and Response Subsystem Using a Rule Engine and a Local Large Language Model
arXiv 2609.26316KV cache Published 2026-09-22
CompKV: Compensation-Aware KV Selection for Long-Context LLM Inference
arXiv 2609.26300failure recovery Published 2026-09-22
Dual-Frontier: When Can an Agent Trust Its World Model?
arXiv 2609.26293failure recovery Published 2026-09-22
FIRE: Failure-Informed Runtime Engineering for Reliable Language-Model Agents
arXiv 2609.26048sparsity and pruning Published 2026-09-22
GeoPair: Geometry-Preserving Cross-Layer Factorization for Training-Free Transformer Compression
arXiv 2609.25963multi-agent orchestration Published 2026-09-22
Governed AI-Agent Coordination for Dementia Care: Architecture, Safety Contracts, and Evidence-Derived Workflow Verification
arXiv 2609.25956quantization Published 2026-09-22
Beyond Scalar Sensitivity: Activation-Aware Mixed-Precision LLM Quantization with Cross-Layer Refinement
arXiv 2609.25916memory and context Published 2026-09-22
When Does Execution Provenance Help Agent Memory Retrieval?
arXiv 2609.25913coding agents Published 2026-09-22
When Should Dependency Updates Invoke Repair Agents? A Lightweight Routing Study
arXiv 2609.25911sparsity and pruning Published 2026-09-22
You Only Need 2/3 of the Chosen Experts: An Empirical Study of Dynamic Expert Pruning in Fine-Grained MoE LLMs
arXiv 2609.25809agent evaluation Published 2026-09-22
The Tasteful Agent: Measuring and Improving Taste in Long-Horizon Tasks
arXiv 2609.25804safety and guardrails Published 2026-09-22
How Strongly Should Task State Influence an LLM Agent?
arXiv 2609.25686tool use Published 2026-09-22
Toolcompass: Guiding Tool Trialing, Not Suppressing It
arXiv 2609.25678agent evaluation Published 2026-09-22
Testing-Driven Reliability Audit of Trajectory-Based Early Outcome Prediction for LLM Agents: Target-Specific Calibration Transfer Persists Within a Single Benchmark
arXiv 2609.25647multi-agent orchestration Published 2026-09-22
Qwen3.8-Omni: Towards Native Omni-Modal Agents
arXiv 2609.25611agent evaluation Published 2026-09-22
Evaluating Coding Agents on Kernel Exploit Generation
arXiv 2609.25591memory and context Published 2026-09-22
AkasicMEM: Governed Enterprise Memory for Agents
arXiv 2609.25563memory and context Published 2026-09-22
Compressing Long Context into Answer-Aligned Memory Embeddings for LLM Inference
arXiv 2609.25537other Published 2026-09-22
Matryoshka attribution: Learning to attribute language model outputs to representations and weights
arXiv 2609.25518other Published 2026-09-21
Universal Fractal Natural Language Decision Map: Real-Time Edge Triage Across Heterogeneous Domains
arXiv 2609.25498serving and batching Published 2026-09-21
Cloud, Edge, or Split? Profiling Onboard and Split Vision-Language Model Deployment for Drone AI
arXiv 2609.25415multi-agent orchestration Published 2026-09-21
Passes Alone, Fails Together: Benchmarking Semantic Coordination in Parallel LLM-Agent Development
arXiv 2609.25396tool use Published 2026-09-21
Extending FunctionGemma for Practical On-Device Mobile Function Calling
arXiv 2609.25373safety and guardrails Published 2026-09-21
From Decorative to Load-Bearing: Task Difficulty Shapes the Causal Role of Chain-of-Thought
arXiv 2609.25366memory and context Published 2026-09-21
Learned Enterprise Data Comprehension: Compression and Routing for Data Agents
arXiv 2609.25286planning and decomposition Published 2026-09-21
When LLM Agents Fail to Read the Room: ReAdapt for Relational Social Reasoning
arXiv 2609.25284tool use Published 2026-09-21
The AI Neuroscientist: An Interactive Agentic Interface for Neuroimaging Analysis
arXiv 2609.25254agent evaluation Published 2026-09-21
Trains but Doesn't Learn: A Post-Training Delivery Benchmark for LLM Agents as Forward-Deployed Engineers
arXiv 2609.25237multi-agent orchestration Published 2026-09-21
Qwen-Audio-Agent Technical Report
arXiv 2609.25195safety and guardrails Published 2026-09-21
Indirect tipping: a social attack surface in AI agent populations
arXiv 2609.25194agent evaluation Published 2026-09-21
FinFIRST: Benchmarking Search Agents for Financial Information Retrieval, Sourcing and Traceability
arXiv 2609.25192tool use Published 2026-09-21
Qwen-Audio-3.1-Realtime: Towards Reliable Agentic Voice Interaction
arXiv 2609.25176other Published 2026-09-21
Who Pays for the KV Cache? Attributing Shared AI Inference Spend Across Kubernetes and LLM Provider Bills
arXiv 2609.24991tool use Published 2026-09-21
Critical-State RL: Diagnosing Trainable States for Multi-Turn Tool Use
arXiv 2609.24985efficient training Published 2026-09-21
onPanda: Efficient Annotation of On-Policy Alignment Data for LLMs and Agents via Token-Level Correction
arXiv 2609.24983efficient training Published 2026-09-21
LoRA-generating hypernetworks for efficient on-device LLM generative personalization
arXiv 2609.24979distillation Published 2026-09-21
Harness-Zero: Harness Distillation via Agent-as-Harness
arXiv 2609.24974other Published 2026-09-21
RRSI: Regularized Recursive Self-Improvement of Agent Harnesses
arXiv 2609.24972agent evaluation Published 2026-09-21
DolphinBench: Mapping the Pareto Frontier of Agent Memory
arXiv 2609.24971safety and guardrails Published 2026-09-21
Emergent Collusion in Long-Horizon LLM Agent Interaction
arXiv 2609.24967agent evaluation Published 2026-09-21
Trajectory-Aware Benchmark Subset Selection for Cost-Efficient Software Engineering Agent Regression Testing
arXiv 2609.24928sparsity and pruning Published 2026-09-21
SLICEChat: Progressive In-Encoder Token Pruning for Whole-Slide Pathology Language Models
arXiv 2609.24894agent evaluation Published 2026-09-21
OSWorld-Pro: Process-based Evaluation for Computer Use Agents
arXiv 2609.24890speculative decoding Published 2026-09-21
SPECTRA: Adaptive Execution of Speculative Decoding on a Runtime-Reconfigurable Tiled Architecture
arXiv 2609.24847agent evaluation Published 2026-09-21
MSI-Bench: Evaluating Multi-Speaker Voice Interaction for Collaborative AI Agents
arXiv 2609.24812safety and guardrails Published 2026-09-21
Decoding Guardrails: XAI-Guided Perturbation Analysis of Prompt Injection Detection
arXiv 2609.24801quantization Published 2026-09-21
When Quantization Preserves Accuracy but Not Evidence: Explanation-Aware Post-Training Quantization for Medical LLMs
arXiv 2609.24799efficient architectures Published 2026-09-21
Complex KDA: Understanding and Enhancing the Expressivity of Kimi Delta Attention
arXiv 2609.24797quantization Published 2026-09-21
NPU Accelerator: Quantized Real-Time Vehicle Detection on PYNQ-Z1 Using FINN
arXiv 2609.24757other Published 2026-09-21
Inference of Unknown Dynamical Components Using Next Generation Reservoir Computing: From Chaotic Systems to Climate Data
arXiv 2609.24754speculative decoding Published 2026-09-21
Adapting Tree-Structured Speculative Decoding to DeepSeek-V4 for Efficient Inference
arXiv 2609.24698efficient training Published 2026-09-21
Muon Can Outperform Dedicated Continual Learning Methods
arXiv 2609.24678other Published 2026-09-21
Circuit Hypernetworks for Quantum-Augmented Diffusion Language Models
arXiv 2609.24657serving and batching Published 2026-09-21
Analytical Power-Aware Provisioning for Prefill-Decode Disaggregated AI Inference
arXiv 2609.24639other Published 2026-09-21
Written as a Record, Read as an Address: What a Forward Pass Leaves in an Operation's KV Cache
arXiv 2609.24635tool use Published 2026-09-21
Ascent: An Agentic System over the Model Context Protocol for Real-World Clinical Data Analysis
arXiv 2609.24620other Published 2026-09-21
Taking a Second Look: Correcting Sea Ice Forecasts with Sparse Observations
arXiv 2609.24591other Published 2026-09-21
Poisson Exchange Beyond Submodularity: Effective Approximation Algorithms for Offline and Online Subset Selection over Matroids
arXiv 2609.24569efficient training Published 2026-09-21
QLoRA Fine-Tuning of Ministral LLM for Sequence-to-Function Protein Annotation
arXiv 2609.24538efficient training Published 2026-09-21
Not All Task Vectors Need Equal Rank: Energy-Proportional Allocation for Model Merging
arXiv 2609.24517safety and guardrails Published 2026-09-21
Beyond Predictable Paths: Redefining AI Security Incident Reporting for Agents
arXiv 2609.24515safety and guardrails Published 2026-09-21
ActGov: Governing LLM Agent Actions via Policy-Constrained Validation
arXiv 2609.24446KV cache Published 2026-09-21
ARM: Attention with Routed-Memory for Learnable Sparse Control
arXiv 2609.24417sparsity and pruning Published 2026-09-21
Artificial Structure Function Search: Preserving Artificial Functional Connectivity for Structured Pruning
arXiv 2609.24401efficient architectures Published 2026-09-21
NAVIR: Neuromorphic Audio-Visual Speech Recognition for Robust Human-Robot Interaction on Edge Hardware
arXiv 2609.24391other Published 2026-09-21
Topographic Training Concentrates Causal Circuits Without Improving Neuron Monosemanticity
arXiv 2609.24379sparsity and pruning Published 2026-09-21
Prescriptive SVD-Inspired Attention via Spectral Energy Retention
arXiv 2609.24370memory and context Published 2026-09-21
Few-Shot Demonstrations Elicit the Use of In-Context World Representations in LLMs
arXiv 2609.24352coding agents Published 2026-09-21
A Lean and Spec-Driven AI-Assisted Software Development Lifecycle for Applied AI Education: The AI-SDLC Approach
arXiv 2609.24348efficient training Published 2026-09-21
A Distributional Optimisation Perspective on Combining Models in Deep Learning
arXiv 2609.24328quantization Published 2026-09-21
The Undetected Damage of Quantization on Retrieval and How to Fix It
arXiv 2609.24322KV cache Published 2026-09-21
KV-COBRA: KV Cache Compression via Co-Optimized Bit-Rank Allocation
arXiv 2609.24298memory and context Published 2026-09-21
TTSE: A Two-Track Online Self-Evolution Framework
arXiv 2609.24289agent evaluation Published 2026-09-21
Canonical Procedural Actions: An Auditable Annotation Protocol for Tool-Use Agent Traces
arXiv 2609.24264memory and context Published 2026-09-21
MemCalib: Benchmarking and Optimizing Memory Use in LLM Agents
arXiv 2609.24259speculative decoding Published 2026-09-21
H-Spec: Parallel Speculative Decoding Without a Drafter-Side KV Cache
arXiv 2609.24197efficient training Published 2026-09-21
CREDO: Variance-Guided Rubric Evolution for Replay-Corrected Credit Assignment
arXiv 2609.24174safety and guardrails Published 2026-09-21
APEXA: Execution-Integrity Enforcement for Multi-Agent LLM Automation of Synchrotron Data Reduction
arXiv 2609.24165tool use Published 2026-09-21
MCP-GRANITE Benchmark: GRANularity Interface TEsting for MCP-Based LLM Agents
arXiv 2609.24161speculative decoding Published 2026-09-21
Acceptance-Aware Draft Model Training for Speculative Decoding
arXiv 2609.24150safety and guardrails Published 2026-09-21
Self-Healing Harness for Runtime Oversight of Agent Self-Modification
arXiv 2609.24130quantization Published 2026-09-21
Q-DEQ: Discrete Solving and Quantization for Deep Equilibrium Models in Time Series Forecasting under Edge Deployment Coding Constraints
arXiv 2609.24042safety and guardrails Published 2026-09-20
FINN-Tro: Exploiting Verification Gaps in Dataflow Inference Accelerators
arXiv 2609.26824quantization Published 2026-09-20
Text Scores Can Miss Waveform Use: A Qwen2-Audio Quantization Case Study
arXiv 2609.26823other Published 2026-09-20
Signal2Symbol: Neuro-Symbolic Temporal Reasoning for Explainable Physiological Time-Series Anomaly Detection
arXiv 2609.26820other Published 2026-09-20
A discrete generative model of neuronal spiking activity on microelectrode arrays
arXiv 2609.23907speculative decoding Published 2026-09-20
GDN Tree-Scan: Served Tree Verification for Recurrent-Hybrid Language Models
arXiv 2609.23900other Published 2026-09-20
Circuit-Diff: Factual Edit-based Intervention Method for Localizing Knowledge in Attribution Graphs
arXiv 2609.23892KV cache Published 2026-09-20
SPLASH: Co-Designing Sparse Attention with High-Bandwidth Flash for Efficient Long-Context Inference
arXiv 2609.23816other Published 2026-09-20
Fast Graph Laplacian Estimation using Effective Resistance
arXiv 2609.23668safety and guardrails Published 2026-09-20
On the Efficiency-Safety Dilemma in Large Reasoning Models
arXiv 2609.23587coding agents Published 2026-09-18
CodeMidas: Scaling Agentic Coding RL Environments from Code Itself
arXiv 2609.22068agent evaluation Published 2026-09-18
RecreationWorld: Scalable and Verifiable Environments for Hybrid Computer-Use Agents
arXiv 2609.22000multi-agent orchestration Published 2026-09-18
Bayesian Belief Layer for Controllable Opinion Dynamics in LLM Agents
arXiv 2609.21997tool use Published 2026-09-18
NemotronLabs VoiceChat: An Open Full-duplex Speech-to-Speech Model with Tool Calling Capabilities
arXiv 2609.21967memory and context Published 2026-09-18
AutoViewMem: Self-Configuring Orthogonal Views for Conversational Long-Term Memory
arXiv 2609.21940agent evaluation Published 2026-09-18
EnterpriseVal: Quantifying the Efficacy, Reliability and Value of Generative AI in the Enterprise
arXiv 2609.21841speculative decoding Published 2026-09-18
RheoSampling: Resolving the One-Hot Dilemma in Stochastic Dynamic-Tree Speculative Decoding
arXiv 2609.21827speculative decoding Published 2026-09-18
SpecQuant: Speculative Decoding with Multi-Parent Quantization for Adaptive LLM Inference
arXiv 2609.21704safety and guardrails Published 2026-09-18
CIPL: A Channel-Aware Framework for Recoverable Privacy Leakage in LLM Agents
arXiv 2609.21686other Published 2026-09-18
Multi-Domain Clustering via Measure Quantization
arXiv 2609.21664other Published 2026-09-18
Weighted Quantum Signal Processing: Low-Depth Polynomial Approximation with Applications to Kolmogorov-Arnold Networks
arXiv 2609.21567agent evaluation Published 2026-09-18
GameLogicBench: Evaluating Coding Agents on Runtime Game Logic with Tick-Level State Assertions
arXiv 2609.21562sparsity and pruning Published 2026-09-18
What Must Survive? Exact Task-Information--State Frontiers for Resource-Sufficient Learning
arXiv 2609.21523safety and guardrails Published 2026-09-18
HE-Guardrail: A Homomorphic Guardrail Against Jailbreak Attacks for Encrypted Large Language Model Inference
arXiv 2609.21484quantization Published 2026-09-18
Understanding LLM Quantization through Activation-Guided Compensation and Orthogonal Residuals
arXiv 2609.21450other Published 2026-09-18
ArenaFlow: From Trajectory Ranking to Hierarchical Credit Propagation for Open-Ended Agent RL
arXiv 2609.21378tool use Published 2026-09-18
What Stops a Small Language Model From Driving a Database Agent
arXiv 2609.21341agent evaluation Published 2026-09-18
Efficient Benchmarking in Production: A Study of an Evolving LLM Agent
arXiv 2609.21267agent evaluation Published 2026-09-18
Verify, Don't Trust: Agentic Model Development for Video Discovery Retrieval at Scale
arXiv 2609.21257KV cache Published 2026-09-18
TierKV: Long-Context On-Device LLMs via Predictive Multi-Tier KV Caching
arXiv 2609.21172other Published 2026-09-17
Can Agents Design Better Chips with a Higher Level Abstraction?
arXiv 2609.21157other Published 2026-09-17
Scaling Forced Alignment to End-User Devices
arXiv 2609.21145quantization Published 2026-09-17
A Multi-Engine Dataflow for MoE Decoding on Scratchpad-Based Tensor Accelerators
arXiv 2609.21137sparsity and pruning Published 2026-09-17
Layerwise Decoupling for Stable Structured Sparsification of Fully Connected Layers
arXiv 2609.21126efficient training Published 2026-09-17
Talk to Me, Jarvis: An Open-Source Edge-Deployable Voice Assistant Framework for Autonomous Racecars
arXiv 2609.21109efficient training Published 2026-09-17
Stiefel-AdamW: Geometry-Aware AdamW for Linear Factorization Blocks
arXiv 2609.21039other Published 2026-09-17
MOSAIC-SR: Transformer-Guided Symbolic Regression for Scientific Equation Recovery
arXiv 2609.20997other Published 2026-09-17
Voice-Light: A Full-Duplex Cascaded Voice Agent with Causal Turn-Taking and Speculative Generation
arXiv 2609.20995planning and decomposition Published 2026-09-17
CaLR: Causal Latent Revision for Robust Diffusion Reasoning
arXiv 2609.20981serving and batching Published 2026-09-17
An Approximate Queueing Model of LLM Inference Serving for SLO-Driven Autoscaling
arXiv 2609.20957safety and guardrails Published 2026-09-17
Coding Agents with an Obstacle-Aware Harness for Safe Robot Manipulation
arXiv 2609.20822agent evaluation Published 2026-09-17
Quantifying Overclaiming Propensity in Frontier LLM Agents
arXiv 2609.20812efficient training Published 2026-09-17
Score Centering Stabilizes Off-policy Reinforcement Learning
arXiv 2609.20807coding agents Published 2026-09-17
An Empirical Study of Harness Design for Coding Agents
arXiv 2609.20804other Published 2026-09-17
Agile-WAM: An Agile Tactile World Action Model for Contact-Rich Robot Control
arXiv 2609.20761attention and kernels Published 2026-09-17
On-Demand Attention: Language Models Know When to Recall
arXiv 2609.20734coding agents Published 2026-09-17
Don't Mask the Environment: Observation Supervision Changes How Agents Explore Under RL
arXiv 2609.20715agent evaluation Published 2026-09-17
Chronicle: Cut-Point Replay for Regression Testing of LLM Agents
arXiv 2609.20625safety and guardrails Published 2026-09-17
Inference-Engine Fingerprinting Attacks are Practical: Exploring Model-Driven Environmental Discovery, Exploitation, and Escape
arXiv 2609.20614agent evaluation Published 2026-09-17
Language-model groups overstate consensus when replaying human deliberation on a reasoning task
arXiv 2609.20543coding agents Published 2026-09-17
SoL-Pi: Recursively Scaling Auto-Research Loops for Efficient Agent Harness
arXiv 2609.20519sparsity and pruning Published 2026-09-17
Truncated automatic sparse differentiation for machine learning interatomic potentials
arXiv 2609.20510sparsity and pruning Published 2026-09-17
Scaling Fourier-Based Sparse Matrix Analysis on GPUs
arXiv 2609.20483planning and decomposition Published 2026-09-17
How Do Agent Harnesses Create Value? Planning Information and Release Control in Stateful LLM Agents
arXiv 2609.20474coding agents Published 2026-09-17
Spotlights: Discovering Improvement Opportunities in Software Repositories
arXiv 2609.20446distillation Published 2026-09-17
Cross-Architecture Foundation-Model Distillation for Edge Flood Segmentation
arXiv 2609.20441speculative decoding Published 2026-09-17
To Copy or Not to Copy: Controlling Speculative Decoding via Intrinsic Model Signals
arXiv 2609.20186efficient training Published 2026-09-17
QUALS: Corpus Equilibrium for Universal Forecasting via Pattern Quantization and Learnability Synchronization
arXiv 2609.20156agent evaluation Published 2026-09-17
MTVA-Bench: Evaluating the Language Model Inside Cascaded Voice Agents
arXiv 2609.20152safety and guardrails Published 2026-09-17
Local Sparsity Enables Unsupervised LLM Safety Detection
arXiv 2609.20129tool use Published 2026-09-17
UnifiedPlayers: Enhance Tool-Integrated Reasoning in Agentic Reinforcement Learning
arXiv 2609.20089tool use Published 2026-09-17
MATCH: Model-Aware Tool Learning with Curriculum Scheduling and Hierarchically Gated Rewards
arXiv 2609.20082efficient training Published 2026-09-17
Marginal utility, matrix factorization, and the Key-Value (KV) cache: a unified information-economic framework for sovereign geo-mining inference
arXiv 2609.20068retrieval and RAG Published 2026-09-17
The Missing Complement: State-Conditioned Minimal Sufficient Evidence for Coding Agents
arXiv 2609.20050KV cache Published 2026-09-17
DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression
arXiv 2609.19969coding agents Published 2026-09-17
Learning and Transferring Closed-Loop Robot Software
arXiv 2609.19906KV cache Published 2026-09-17
D-Quant: Driftable Entropy Coding for KV Cache Quantization
arXiv 2609.19880speculative decoding Published 2026-09-17
Zarya: A Hybrid Autoregressive--Masked Diffusion Language Model with Flexible Training and Dual-Mode Inference
arXiv 2609.19868efficient training Published 2026-09-17
Dual-Axis Policy Optimization for LLM Agents: Bayesian Feedback Attribution and Trajectory Mass Normalization
arXiv 2609.19830multi-agent orchestration Published 2026-09-17
Rethinking Multi-Agent Collaboration: When More Is Less
arXiv 2609.19759efficient training Published 2026-09-17
AutoData: Agentic Search for Pre-training Data Selection
arXiv 2609.19754failure recovery Published 2026-09-17
Syndrome Decoding for Silent Data Corruption in Quantized Integer GPU Arithmetic
arXiv 2609.19743memory and context Published 2026-09-17
LearnActCoder: Role-Aware Error Memory for Adaptive Clinical Coding Agents
arXiv 2609.19721safety and guardrails Published 2026-09-17
SoK: Trading Agents or Market Crashers? Dissecting Robustness and Security Failures in Academic Financial LLM Trading Schemes
arXiv 2609.19705attention and kernels Published 2026-09-17
Understanding and Exploiting Diagonal Attention Sparsity in Autoregressive Image Generation
arXiv 2609.19702quantization Published 2026-09-17
MiX: Micro-Inverted-Scaling for End-to-End Low-Bit Vision-Language Model Acceleration
arXiv 2609.19683other Published 2026-09-17
CoRe: Coherence and Relational Alignment for Multivariate Time Series Forecasting
arXiv 2609.19670KV cache Published 2026-09-17
PrefixBench-H100: Characterizing Prefix Reuse and Time-to-First-Token in H100 LLM Serving
arXiv 2609.19657retrieval and RAG Published 2026-09-17
Self-Evolving Search Index
arXiv 2609.19656safety and guardrails Published 2026-09-17
From Intent to Action: Benchmarking LLM Safety in Vehicle Voice Command Authorization
arXiv 2609.19630agent evaluation Published 2026-09-17
DeltaSelect: Affordable A/B Testing for Coding Agents
arXiv 2609.19607safety and guardrails Published 2026-09-17
Red-Teaming Auto Mode: Improving Blocking Classifiers Against Malign Coding Agents
arXiv 2609.19587coding agents Published 2026-09-17
Self Improvement via Fast Tree-search
arXiv 2609.19526agent evaluation Published 2026-09-17
A Unified Evaluation Framework for Trustworthy Large Language Models, Agentic AI, and Multimodal Systems
arXiv 2609.19524attention and kernels Published 2026-09-16
Elastic Threshold Attention: Learned Contextual Sparsity for Long-Context Decoding
arXiv 2609.20888other Published 2026-09-16
Sparse Priors for Efficient Distribution Learning
arXiv 2609.20883memory and context Published 2026-09-16
Reputation as Community Memory for the Agentic Web
arXiv 2609.19502safety and guardrails Published 2026-09-16
Safety Beyond the Interface: Detecting Harm via Latent States in Large Language Models
arXiv 2609.19472efficient training Published 2026-09-16
Compositional Reasoning in Language Models under Reinforcement Learning Post-Training
arXiv 2609.19465safety and guardrails Published 2026-09-16
Closed-World Resolution Against Tool Hallucination in LLM Agents
arXiv 2609.19425safety and guardrails Published 2026-09-16
MAGS: Multi-agent Auto-formalization Guarantees Safety for Agentic Outputs
arXiv 2609.19391other Published 2026-09-16
FCx: An algorithm for finding Feasible Counterfactual Explanations
arXiv 2609.19383tool use Published 2026-09-16
A frontend-backend architecture for tool calls in full-duplex speech models
arXiv 2609.19334agent evaluation Published 2026-09-16
Characterizing Web Search by Conversational LLM Agents: From Search Decisions and Strategies to Results and Responses
arXiv 2609.19244other Published 2026-09-16
Randomized SVD Approximations for Spectral Co-Clustering of Word-Document Matrices
arXiv 2609.19243efficient training Published 2026-09-16
Block Parallelism For Efficient Distributed Long-Context Diffusion Language Model Training
arXiv 2609.19242safety and guardrails Published 2026-09-16
PAPC: Platform Mediation for Privacy-Propagation Externalities in AI-Mediated Workflows
arXiv 2609.19226distillation Published 2026-09-16
Layer-wise Curriculum Learning for Efficient LLM Compression
arXiv 2609.19213KV cache Published 2026-09-16
MeshKV: A Network-on-Chip KV Cache Fabric for Scalable Transformer Decoding Accelerators
arXiv 2609.19207multi-agent orchestration Published 2026-09-16
Position: It is Time to Virtualize Foundation Models with a Self-evolving Operating System Layer
arXiv 2609.19203other Published 2026-09-16
Objective vs. Search: Decomposing What Makes a Good Tokeniser
arXiv 2609.19145efficient training Published 2026-09-16
A Zeroth-Order Paradigm for LLM Preference Alignment
arXiv 2609.19144tool use Published 2026-09-16
Affora: A Design System for Agent-Friendly Interfaces
arXiv 2609.19125efficient architectures Published 2026-09-16
Training-Adaptive Convolutional Sparse Coding via Information Bottleneck for Robust Visual Representation
arXiv 2609.19122other Published 2026-09-16
A General Kernel Framework for Non-CND Distance Measures Using |D|-Dimensional Sparse Landmark Embeddings
arXiv 2609.19083other Published 2026-09-16
Probabilistic Linear Explanations
arXiv 2609.19077memory and context Published 2026-09-16
Long-Lived Characters, Local Inference: Incremental Memory Maintenance for Game NPCs
arXiv 2609.18935sparsity and pruning Published 2026-09-16
Higher-order pruning of experts in mixture-of-experts language models
arXiv 2609.18916agent evaluation Published 2026-09-16
Beyond Outcomes: Dual-View Relational Learning for Efficient Agent Benchmarking
arXiv 2609.18909agent evaluation Published 2026-09-16
ASLEval: Measuring Privacy Exposure Displacement in LLM Agent Sessions
arXiv 2609.18864serving and batching Published 2026-09-16
Ask the Tool, Don't Guess: Agent Tool Calls Hold Their Progress, and the Serving System Should Read It
arXiv 2609.18849agent evaluation Published 2026-09-16
ReFigBench: Benchmarking Scientific Figure Reconstruction as Editable PowerPoint Artifacts
arXiv 2609.18844memory and context Published 2026-09-16
Infinite-Parameter LLMs: Generating and Adapting Weights from Live Data
arXiv 2609.18842agent evaluation Published 2026-09-16
ProgramDistill: From Interactive Web Apps to Verifiable Reference-Guided SWE Tasks
arXiv 2609.18805other Published 2026-09-16
Quantifying the Effect of HCLs on a Fixed-Microarchitecture MXFP4 Accelerator
arXiv 2609.18792other Published 2026-09-16
Beyond Truncation: Rethinking LLM Decoding as Ensemble Pruning
arXiv 2609.18723memory and context Published 2026-09-16
HBFlex: A Flexible Memory System for Bridging Fine-Grained LLM States and Coarse-Grained HBF Parallel Execution
arXiv 2609.18675agent evaluation Published 2026-09-16
PACT: Can Enterprise AI Assistants Be Trusted Under Pressure?
arXiv 2609.18605agent evaluation Published 2026-09-16
Recursive Reasoning or Statistical Extrapolation? In-Context Learning in Multi-Agent Interdependent Decision-Making
arXiv 2609.18591other Published 2026-09-16
Provable Guarantees and Efficient Learning of Structural Equation Models with Latent Confounders
arXiv 2609.18535other Published 2026-09-16
Machine Translation between English and Syriac (East Syriac Dialect) using Statistical Machine Learning
arXiv 2609.18529multi-agent orchestration Published 2026-09-16
AeroWeaver: An Embodied-Agent Harness for Weaving Aerial Skills into Distributed, Adaptive Swarm Execution
arXiv 2609.18520other Published 2026-09-16
ActionPiece: Rethinking Action Tokenization for Autoregressive Vision-Language-Action Models
arXiv 2609.18487agent evaluation Published 2026-09-16
The Mirage of Calibrated Confidence: Trajectory-Independence of Verbalized Confidence in Vision-Language Models
arXiv 2609.18453retrieval and RAG Published 2026-09-16
M-SQE: Multilingual Skill Quality Estimation for Enhancing Language Equality in Agentic Skill Use
arXiv 2609.18445coding agents Published 2026-09-16
WetRobo: A Reproducible Robot Kit for Coding Agents in Biological Laboratories
arXiv 2609.18435efficient training Published 2026-09-16
Dependency-Aware Trajectory Refinement for Efficient Multi-Turn Agent Fine-Tuning
arXiv 2609.18417efficient training Published 2026-09-16
Gradient Descent with Stochastic Subspaces via Persistence of Memory
arXiv 2609.18416agent evaluation Published 2026-09-16
Bad Genius: Counterfactual-Guided Harness Evolution Beyond Task-Specific Shortcuts
arXiv 2609.18366other Published 2026-09-16
RecMorph: Topology-Guided Spatial Recurrence for Generalized Morphology Control
arXiv 2609.18359failure recovery Published 2026-09-16
Rollback the World, Keep the Reflection: Rollback-Induced Reflection for Long-Horizon LLM Agents
arXiv 2609.18304safety and guardrails Published 2026-09-16
Who Audits Whom, on What Substrate, with What Evidence? An Independence-Graded Audit Protocol for Agentic AI
arXiv 2609.18272other Published 2026-09-16
${M}^2$Tok: Multi-head Multi-codebook Discrete Action Tokenization for Vision-Language-Action Models
arXiv 2609.18259other Published 2026-09-16
Reinforcement Learning for Real-Time Vision-Language-Action Policies
arXiv 2609.18207memory and context Published 2026-09-16
WFM: Wiki Foundation Model for Complex Agentic Reasoning
arXiv 2609.18182multi-agent orchestration Published 2026-09-16
DualSQL: Text-to-SQL with Multi-Agent Reinforcement Learning
arXiv 2609.18135quantization Published 2026-09-16
Colla-Q: Toward Collaborative Experts in MoE Quantization via Minimax Precision Balancing
arXiv 2609.18131safety and guardrails Published 2026-09-16
Symbolic Temporal Supervision of LLM Agents Using Contracts
arXiv 2609.18128serving and batching Published 2026-09-16
Token Latency Fairness: Performance Isolation for Multi-Tenant LLM Serving
arXiv 2609.18112serving and batching Published 2026-09-16
SSD-LLaMA: SSD-Native Inference for Trillion-Parameter MoE at 1+ Token/s on a Consumer PC
arXiv 2609.18110multi-agent orchestration Published 2026-09-16
Agora: Git as Shared Memory for Collective AutoResearch
arXiv 2609.18094serving and batching Published 2026-09-16
vidax: A Unified JAX Framework for Video Generative Models on Accelerator Meshes
arXiv 2609.18077agent evaluation Published 2026-09-16
A Calibrated Instrument for Measuring How Inference Optimizations Affect Output Quality
arXiv 2609.18005safety and guardrails Published 2026-09-16
QuanText: Protecting Dataset-Level Secrets in Textual Data Sharing
arXiv 2609.17995safety and guardrails Published 2026-09-16
Whom Do AI Agents Work For? Role Assignment Induces Sponsorship Bias in LLM Recommenders
arXiv 2609.17989agent evaluation Published 2026-09-16
RideWay: Benchmarking Efficient Task Completion for Tool-Using Language Agents
arXiv 2609.17985tool use Published 2026-09-16
TuiML: Machine Learning for AI Agents
arXiv 2609.17984efficient training Published 2026-09-16
Encoder Awakening via Adapters: Effective Domain-Adaptive Fine-tuning of Speech-LLMs
arXiv 2609.17981speculative decoding Published 2026-09-16
ASPIRE: Asynchronous Batched Self-Speculative Decoding for Long-Context LLM Inference
arXiv 2609.17943agent evaluation Published 2026-09-15
ERPBench: A State-Grounded Evaluation Paradigm for Computer-Use Agents in Enterprise Software
arXiv 2609.17885tool use Published 2026-09-15
SFT or RL for Tool-Calling Agents? A Controlled Study Across Data, Method, and Scale
arXiv 2609.17848safety and guardrails Published 2026-09-15
Reflections on Trusting Trust, Revisited: Contaminating Self-Modifying AI Coding Agents with Poisoned Benchmarks
arXiv 2609.17817other Published 2026-09-15
Principled Koopman Representations with Kalman Inference for Efficient Time-Series Prediction
arXiv 2609.17815sparsity and pruning Published 2026-09-15
FairCompressAgent: An Agentic Framework for Fairness-Aware Model Compression for FPGA Deployment
arXiv 2609.17786efficient training Published 2026-09-15
Derivative-Free Structured Updates for Muon
arXiv 2609.17759agent evaluation Published 2026-09-15
A Large-Scale Empirical Study of Quality Assurance Practices and Gaps in AI Agents
arXiv 2609.17698agent evaluation Published 2026-09-15
GraphEcho: Structural Redundancy and Evidence Provenance in LLM Graph Agents
arXiv 2609.17695KV cache Published 2026-09-15
Fathom: Per-Query Read Depth for Sparse Decoding over Offloaded KV Caches
arXiv 2609.17652tool use Published 2026-09-15
EvolveTrade: Experience-Driven Policy Refinement for Self-Evolving LLM Trading Agents
arXiv 2609.17632sparsity and pruning Published 2026-09-15
What Breaks Under Pruning in Smart Homes, and When? Evaluating LLM Degradation Across Architectures and Task Complexity
arXiv 2609.17515efficient architectures Published 2026-09-15
LACE: Layer-Wise Compression for Dynamic Frame Rate Codecs
arXiv 2609.17509serving and batching Published 2026-09-15
JustFit: 200K-Token LLM Serving on a 24 GiB Laptop with Just-in-Time State Management
arXiv 2609.17475multi-agent orchestration Published 2026-09-15
Decomposition Buys Integrity, Not Yield
arXiv 2609.17464other Published 2026-09-15
Right Tool, Right Job: Native-Language Evaluation, Tokenizer Sensitivity, and Methodological Findings from a French-Only BabyLM
arXiv 2609.17435agent evaluation Published 2026-09-15
Coding Agents Have Converged: Why the SWE-bench Leaderboard Can No Longer Order Its Top Entries, and What to Measure Instead
arXiv 2609.17394coding agents Published 2026-09-15
FlashVector: Agent for Hierarchical Model Serving Stack Optimization
arXiv 2609.17391memory and context Published 2026-09-15
Where Should a Document Live: Context, Representations, or Parameters?
arXiv 2609.17346multi-agent orchestration Published 2026-09-15
Self-Emergence Agent Architecture:Behavior-Inertia HMM, Reflexive Metacognition,and Social-Contrastive Self-Modeling
arXiv 2609.17331agent evaluation Published 2026-09-15
Emergence World: Adversarial Stress-Testing of Long-Horizon Multi-Agent Systems
arXiv 2609.17320speculative decoding Published 2026-09-15
ECHO: Early-layer Collaborative Hierarchical Orchestration with Bonus Logits in Speculative Decoding
arXiv 2609.17241speculative decoding Published 2026-09-15
LoopSpec: Pipelined Self-Speculative Decoding for Looped Transformers
arXiv 2609.17184other Published 2026-09-15
A unified framework for global and local interpretability using adaptive derivative-ordered random explanation
arXiv 2609.17171KV cache Published 2026-09-15
Shared-Prefix KV Reuse Across Standard LoRA Adapters: Quality and Serving Tradeoffs
arXiv 2609.17109tool use Published 2026-09-15
Symbolic Separation: Grounding Deep Agents in Knowledge Graphs for Trustworthy Operational Data Analytics
arXiv 2609.17107other Published 2026-09-15
Budgeted Express-Mesh: Traffic-Aware Link Placement and Deadlock-Free Adaptive Routing
arXiv 2609.17057other Published 2026-09-15
Near-Optimal Nonconvex Matrix Completion
arXiv 2609.17048other Published 2026-09-15
Learning Options for Compositional Motor Control with Adapter Banks
arXiv 2609.17042sparsity and pruning Published 2026-09-15
CLARE: Scalable Class-Incremental Continual Learning via a Sparsity-Based Framework
arXiv 2609.17026coding agents Published 2026-09-15
RepoAtlas: Guiding Coding Agents via Evolving Multimodal Repository Views
arXiv 2609.16936other Published 2026-09-15
Causal Discovery via Transformed Low-Rank Quantile Surfaces
arXiv 2609.16931memory and context Published 2026-09-15
TEMPO: Learning Temporal Context for Dynamic Robot Manipulation
arXiv 2609.16864other Published 2026-09-15
LCAP: Population-Informed Latent Chip Adaptation from Few Output Probes for Photonic Neural Networks
arXiv 2609.16823other Published 2026-09-15
Noise2Noise Revisited: Training Pair Distributions Dominate Loss Choice in Self-Supervised Denoising
arXiv 2609.16788efficient training Published 2026-09-15
Turn-level Multiscale Density Ratio Estimation for LLM Agents
arXiv 2609.16760speculative decoding Published 2026-09-15
GrowMTP: Can RL Grow Its Own Draft Head?
arXiv 2609.16648memory and context Published 2026-09-15
EchoPath: Execution-Level Replayable Memory for GUI Agents
arXiv 2609.16635KV cache Published 2026-09-15
Divergence Timing and Cumulative Disagreement under KV-Cache Eviction
arXiv 2609.16617other Published 2026-09-15
Certified Inference and Training for Deep Equilibrium Networks: A Continuation Framework with Polynomial Complexity Guarantees
arXiv 2609.16485memory and context Published 2026-09-15
Protocol-Preserving Context Trimming for Agentic Workflows: Benefits, Failure Regimes, and Budget Guardrails
arXiv 2609.16461speculative decoding Published 2026-09-15
Early-Bird Decoding: Accelerating Diffusion LLMs with Learnable Block Sizes and Parallel Sampling
arXiv 2609.16450other Published 2026-09-15
Adaptive Bayesian Partner Selection for Federated Clinical Centers
arXiv 2609.16446agent evaluation Published 2026-09-14
AgentPProf: Semantic Profiler for Long Horizon AI Agents
arXiv 2609.20301other Published 2026-09-14
Interpreting and Steering LLM Agents for Social Simulations
arXiv 2609.16436safety and guardrails Published 2026-09-14
Implementing a White-Box Undetectable Backdoor for Random Fourier Features
arXiv 2609.16403quantization Published 2026-09-14
Where Post-Training Quantization Breaks Text Embedders: A Measured Map Across Four Embedder Families
arXiv 2609.16391efficient architectures Published 2026-09-14
FINNAS: FINN-Guided Hardware-Aware NAS and Pruning for FPGA Jet Substructure Classification
arXiv 2609.16367other Published 2026-09-14
FSNIC: A Low-Latency Flow-Based Intrusion Detection Architecture for FPGA SmartNICs
arXiv 2609.16363other Published 2026-09-14
EBL: Efficient Broad Learning for Distributed Adaptive Harmonic Analysis
arXiv 2609.16358agent evaluation Published 2026-09-14
BLINDSPOT: A Benchmark for Safety and Refusal Calibration in Long-Horizon Tool-Using Agents
arXiv 2609.16305coding agents Published 2026-09-14
Assurance Envelopes for Autonomous Coding Agents: Minimum-Cost Evidence for Software Change
arXiv 2609.16302safety and guardrails Published 2026-09-14
AgentGuard: Learning Execution Guardrails from Anomalous Coding-Agent Trajectories
arXiv 2609.16287tool use Published 2026-09-14
Spurious Tool Use: When RL Agents Learn the Wrong Reason to Act
arXiv 2609.16268agent evaluation Published 2026-09-14
CADWorld: Computer-Use Benchmark for Long-Horizon Computer-Aided Design
arXiv 2609.16251safety and guardrails Published 2026-09-14
Artificial intelligence and biosecurity: capabilities, threat pathways, and defense-in-depth governance
arXiv 2609.16213safety and guardrails Published 2026-09-14
Decoy Direction Optimization: A Post-Hoc Defense Against LLM Abliteration
arXiv 2609.16204efficient architectures Published 2026-09-14
Anatomy of Associative Recall in Fixed-State Recurrences: A Matched-State Decomposition, an Interference Wall, and a Curriculum That Breaks It
arXiv 2609.16183other Published 2026-09-14
Skeletal Prototypes on Iterative Nerve Expansions
arXiv 2609.16170KV cache Published 2026-09-14
LLM Inference in a Flash!
arXiv 2609.16161safety and guardrails Published 2026-09-14
Universal Defenses for Tool-Integrated LLM Agents Against Adversarial Attacks
arXiv 2609.16098coding agents Published 2026-09-14
Coaching Qwen3 Coder 30B to Think Like a CodeClash Arena Agent
arXiv 2609.16096distillation Published 2026-09-14
Distilling Foundation Models for Agentic What-If Reasoning:Cost, Latency, and Governance in a Hybrid LLM+SLM Architecture
arXiv 2609.16091quantization Published 2026-09-14
Is INT8 Portable? A Cross-Platform Measurement Study of Quantized Inference on Embedded and Automotive Accelerators
arXiv 2609.16085tool use Published 2026-09-14
The Router Within: Eliciting Native Skill Routing from a Frozen LLM
arXiv 2609.15982multi-agent orchestration Published 2026-09-14
Discovery Foundation Models: Toward Open-Ended Discovery Intelligence
arXiv 2609.15973safety and guardrails Published 2026-09-14
Privacy-Aligned Personalized Federated Learning with Compact Adaptation and Variable-Length Gaussian Communication
arXiv 2609.15950multi-agent orchestration Published 2026-09-14
HypoEvolve: Genetic Algorithms Enable Multi-Agent LLMs to Discover Scientific Hypotheses
arXiv 2609.15938coding agents Published 2026-09-14
Using Agentic AI for contextualized and multifaceted code review at Ericsson
arXiv 2609.15877efficient training Published 2026-09-14
LLM-Based Schema-Aware Split Learning for Privacy-Preserving Mental Distress Prediction Across Heterogeneous Surveys
arXiv 2609.15871tool use Published 2026-09-14
EvoOntology: A Self-Evolving Ontology Layer for Data Agents
arXiv 2609.15779other Published 2026-09-14
Bench2Dex: Benchmarking Visuo-Tactile Bimanual Dexterous Manipulation Across Dexterous Hands
arXiv 2609.15726efficient training Published 2026-09-14
Principal-timestep Restricted Init via Sparse Matrix-decomposition in Flow-matching
arXiv 2609.15643attention and kernels Published 2026-09-14
DeepSeek-V4-Flash on AMD gfx90a: Correctness Recovery and Inference Performance Engineering
arXiv 2609.15627other Published 2026-09-14
Beyond AI Literacy: A Structured Review and Exploratory Meta-Analysis of Measures for Competent Generative-AI Use
arXiv 2609.15624other Published 2026-09-14
Where to Compute and How to Interact: Operator-Readable Adaptation with Gauge-Aware Transport
arXiv 2609.15620efficient training Published 2026-09-14
Option-Aware Retrieval and Task-Specific VLM Adaptation for Medical VQA
arXiv 2609.15530efficient architectures Published 2026-09-14
To Each Language Its Tokenizer: Modular Tokenizers for Efficient Multilingual LLMs
arXiv 2609.15528speculative decoding Published 2026-09-14
How Lossless Is Lossless Speculative Decoding? The Role of Numerical Precision in Orthrus
arXiv 2609.15504agent evaluation Published 2026-09-14
Turkish MMLU Pro: Traceable Option Augmentation and Its Validity Limits in Turkish Multiple-Choice Evaluation
arXiv 2609.15467memory and context Published 2026-09-14
MarKey: Marginal Utility Guided Greedy Keyframe Selection for Long Video Understanding
arXiv 2609.15408agent evaluation Published 2026-09-14
Clean Scores, Buried Evidence, and Confident Wrong: A Receipt-Based Audit of Frontier Agentic QA
arXiv 2609.15319agent evaluation Published 2026-09-14
When Agents Slow Down: Understanding LLM Agents' Test-Time Strategies via Elo-per-token Analysis
arXiv 2609.15309retrieval and RAG Published 2026-09-14
Reason What Matters: Retrieval-Grounded Reasoning for Universal Multimodal Embeddings
arXiv 2609.15296safety and guardrails Published 2026-09-14
Why LLM Agents Collapse Without Oversight: The Enforcement Gap as the Mechanism Behind Emergence World Failures
arXiv 2609.15293multi-agent orchestration Published 2026-09-14
MUSE: A Theory-Harnessed Story Engine for Vibe Narrativizing
arXiv 2609.15188efficient training Published 2026-09-14
PACE: Progressive Angular-to-Norm Contrastive Embedding
arXiv 2609.15152sparsity and pruning Published 2026-09-14
SparseTalk - Sparsifying 3D Gaussian Language Fields for Efficient 3D Visual Question Answering
arXiv 2609.15137safety and guardrails Published 2026-09-14
HazardAuditor: From Executable Threats to Safer Computer-Use Agents
arXiv 2609.15134coding agents Published 2026-09-14
OpenAI4S: Code as Action, Science as Sessions
arXiv 2609.15096multi-agent orchestration Published 2026-09-14
Translating the Translator: Decomposing the Cost of English-Forced Inter-Agent Communication
arXiv 2609.15079tool use Published 2026-09-14
Salesforce Koa: An Enterprise Language Model for Agentic Tool Use
arXiv 2609.15066multi-agent orchestration Published 2026-09-14
BusMA: A Bus Communication Substrate for Multi-Agent Systems
arXiv 2609.15054safety and guardrails Published 2026-09-14
SpliTEE: Improving LLM Inference on Trusted Hardware with Differentially Private GPU Outsourcing
arXiv 2609.15039efficient training Published 2026-09-14
MoARa: Module-Aware Rank Allocation and Structure-Preserving Decomposition for Low-Rank LLM Pre-training
arXiv 2609.15037KV cache Published 2026-09-14
Validating Hybrid-State Cache Recovery for GLM-5.3-Flash with vLLM and LMCache
arXiv 2609.15030safety and guardrails Published 2026-09-14
Overflip: Repetition-Induced Label Flips in Guardrail Models
arXiv 2609.15013safety and guardrails Published 2026-09-14
Semantic-TVM: Structure-Preserving Trustworthy Virtual Memory for Memory-Augmented and Tool-Using Agents
arXiv 2609.15011agent evaluation Published 2026-09-14
MTAC-IFBench: Benchmarking Instruction-Following in Multi-Turn Agentic Coding
arXiv 2609.14992safety and guardrails Published 2026-09-14
ActGuard: Pre-execution Action Auditing against Indirect Prompt Injection in LLM Agents
arXiv 2609.14987multi-agent orchestration Published 2026-09-14
Converting Sequenced Fuzzy Cognitive Maps to Causal Virtual Worlds with Large Video Generators
arXiv 2609.14985KV cache Published 2026-09-14
AgentKV: Phase-Aware KV Eviction for Agentic LLMs
arXiv 2609.14872quantization Published 2026-09-14
GGUF-Metadata Prediction of Single-Sequence llama.cpp Throughput Across Three Systems
arXiv 2609.14864speculative decoding Published 2026-09-13
Self-Orchestrating Language Models: Leveraging Semantic Dependence for Efficient Inference
arXiv 2609.14850other Published 2026-09-13
A Functional SVD Framework for Regularized Multivariate Functional PCA with Dual Penalization
arXiv 2609.14815agent evaluation Published 2026-09-13
Mind Which Bird You Favour: Parameterizing Adequacy-Fluency Balance in Meta-Evaluation of Machine Translation
arXiv 2609.14795retrieval and RAG Published 2026-09-13
TriCalRAG: A Three-Strategy, Retrieval-Augmented Benchmark for On-Premise LLM-Based Root Cause Analysis in AIOps
arXiv 2609.14762safety and guardrails Published 2026-09-13
Refusal Reads Only a Slice of What the Model Knows: Harm-Keyed Routing and Its Exceptions Across Model Families
arXiv 2609.14759quantization Published 2026-09-13
WaterKron and FlipFlop Hessian: Information-Theoretically Grounded Quantization with Kronecker-factored Hessians
arXiv 2609.14706other Published 2026-09-12
Entropy-Punctured Bloom Filters for Memory-Efficient Machine Learning
arXiv 2609.14187serving and batching Published 2026-09-12
When Faster VLA Deployment Changes Closed-Loop Behavior: Task Success-Latency Analysis of SmolVLA Across PyTorch and ONNX Variants
arXiv 2609.14146other Published 2026-09-12
Hardware-Aware Learned Representation Compression for Distributed In-Sensor Vision
arXiv 2609.13947other Published 2026-09-11
One Spectrum, Two Resources: Data-Memory Scaling in Autoregressive Prediction
arXiv 2609.13500efficient architectures Published 2026-09-11
Kraken: LLM-based Speech-to-Speech Translation via Low-bitrate VQ and Dual-path Source Conditioning
arXiv 2609.13045quantization Published 2026-09-11
Attention Quantization for Tabular Foundation Models
arXiv 2609.13031attention and kernels Published 2026-09-11
Dissecting GPU Utilization for LLM Inference on Nvidia Hopper
arXiv 2609.12923serving and batching Published 2026-09-11
Quality-Constrained Routing over a Fixed Pool of Quantized Mixture-of-Experts Instances
arXiv 2609.12550other Published 2026-09-11
HoliBench: A Cross-Platform Benchmarking and Deployment Toolkit for Foundation Models in CPS-IoT Applications
arXiv 2609.12412sparsity and pruning Published 2026-09-11
ESTS at WMT26: Routing-Informed Expert Pruning for Model Compression
arXiv 2609.12310quantization Published 2026-09-10
Vortex: Bridging Extreme Compression and Efficient LLM Inference
arXiv 2609.12208quantization Published 2026-09-10
Why Does Post-Training Quantization Work?
arXiv 2609.11716quantization Published 2026-09-10
Structured Transforms for Low-Overhead Quantization of Language Models
arXiv 2609.11687memory and context Published 2026-09-10
Grounding Agent Memory: Environment-Probing Curation for Enterprise Agents
arXiv 2609.11060speculative decoding Published 2026-09-09
PELM: Power Efficient On-Device LLM Inference with Speculative Decoding and Dynamic Voltage Frequency Scaling
arXiv 2609.09662speculative decoding Published 2026-09-08
Osprey: Target-agnostic Pre-training Makes Stronger Drafters in Speculative Decoding
arXiv 2609.09338memory and context Published 2026-09-08
Procedural Graphs: Self-Evolving Execution Structures for LLM Agents
arXiv 2609.09153memory and context Published 2026-09-08
Closing the Consistency Gap: Self-Evolving Agents That Learn to Stay on Course
arXiv 2609.08832KV cache Published 2026-09-07
MetaKV: Adaptive KV Cache Compression for Constrained LLM Inference
arXiv 2609.07966sparsity and pruning Published 2026-09-07
Train Overcomplete, Deploy Compact: Scaling Recovery Capacity for Structured LLM Pruning
arXiv 2609.06974safety and guardrails Published 2026-09-07
AgentDrift: A Step-Labeled Benchmark of Injection-Hijacked LLM Agent Trajectories
arXiv 2609.06972agent evaluation Published 2026-09-05
Explaining AI Agents Through Execution Traces
arXiv 2609.06063multi-agent orchestration Published 2026-09-04
Inference-Time Graph Engineering for Multi-Agent LLM Workflows
arXiv 2609.05774tool use Published 2026-09-04
Beyond Prompts: Measuring and Optimizing LLM Tool-Agent Harnesses
arXiv 2609.05736sparsity and pruning Published 2026-09-04
Deep Microcompression: Structured Pruning and Bit-packed Quantization for Microcontrollers
arXiv 2609.05081planning and decomposition Published 2026-09-04
TROVE: Adaptive Agent Skill Orchestration via Trace-Grounded Route Validation and Editing
arXiv 2609.05019agent evaluation Published 2026-09-03
Harbor Adapters and Harbor-Index: Infrastructure and a Curated Meta-Dataset for Large-Scale Agentic Evaluation
arXiv 2609.04298agent evaluation Published 2026-09-03
EVOHARNESSBENCH: Can Your Agents Keep Pace with an Evolving Harness?
arXiv 2609.04280KV cache Published 2026-09-03
Random Attention: Rethinking KV Cache Eviction for Efficient Reasoning
arXiv 2609.03430KV cache Published 2026-09-02
Quality Recovery for Quantized KV Caches via Low-Rank Attention Adaptation
arXiv 2609.04263quantization Published 2026-09-02
Unfolding the Leech Lattice: Fused Multi-Shell Decoding and VRAM Layouts for 2-Bit LLM Weights
arXiv 2609.02652tool use Published 2026-09-01
Harness Engineering in LLM Tool Use via Agent-Native Reusable Tool Primitives
arXiv 2609.01736serving and batching Published 2026-09-01
OUTLETS: Output-Length Prediction from Speculative Decoding Backbones
arXiv 2609.01068memory and context Published 2026-09-01
ContextPipe: Database-Inspired Context Assembly for Long-Horizon Agents
arXiv 2609.00749quantization Published 2026-08-31
HBQ: Hierarchical Scaling Block Quantization with Hardware-Efficiency-Aware Design for Accurate LLM Inference
arXiv 2609.00450attention and kernels Published 2026-08-31
Faster Than Flash: Exploiting Attention Sparsity for Efficient Long-Context Decoding
arXiv 2609.00097memory and context Published 2026-08-31
Measure Before You Manage: Evaluating Agent Working Memory in Coding Agents
arXiv 2608.31057coding agents Published 2026-08-31
A Phased Workflow for Operating LLM-Based Coding Agents
arXiv 2608.30701speculative decoding Published 2026-08-31
Strong Drafts Need Compact Memories: Long-Context Speculative Decoding with Compressed KV Cache
arXiv 2608.30252tool use Published 2026-08-31
CAST: Critique-Aware Supervision for Training Reliable Long-Horizon Tool-Calling Agents
arXiv 2608.30147quantization Published 2026-08-30
Budget-Aware Compression Pipeline for Single-GPU LLM Inference: Methods, Trade-offs, and Coupling Effects
arXiv 2608.30076safety and guardrails Published 2026-08-30
Reachability-Based Capability Confinement for LLM Agents under Indirect Prompt Injection
arXiv 2608.30041KV cache Published 2026-08-28
SemKV: Semantic Mixed-Precision KV Cache Quantization Guided by the Quality Cliff for Long-Context LLM Inference
arXiv 2608.28911quantization Published 2026-08-28
A Method for Layer Bit-Width Allocation in LLM Quantization via Performance Maximization Under a Quality-Degradation Constraint
arXiv 2608.28003coding agents Published 2026-08-28
openJiuwen: Beyond Static Harnesses for Long-Horizon Coding Agents
arXiv 2608.27969other Published 2026-08-27
Stochastic Estimation of Transduced Language Models
arXiv 2608.27428agent evaluation Published 2026-08-27
DuMateBench: Evaluating Autonomous Agents in Complex Real-World Workflows
arXiv 2608.26546multi-agent orchestration Published 2026-08-27
Zero-Shot Self-Orchestration with Ledger-Based Control for Improved LLM Coding Performance
arXiv 2608.26480speculative decoding Published 2026-08-26
AsymSpec: Context-Asymmetric Speculative Decoding for Agentic LLMs
arXiv 2608.26004other Published 2026-08-25
Hydra: Phase-Aware Workload Characterization of LLM Inference across Edge SoC Generations, Backends, and Quantization Levels
arXiv 2608.25053tool use Published 2026-08-25
Joint Optimization of Tool Creation and Use for Large Language Model Agents
arXiv 2608.24571agent evaluation Published 2026-08-25
PeakBench: Benchmarking Resource-Aware Tool Invocation in LLM Agents
arXiv 2608.24509sparsity and pruning Published 2026-08-25
Compression Trinity: Exploring Sparsity, Quantization, and Low-Rank Approximations for LLM Compression
arXiv 2608.24070speculative decoding Published 2026-08-25
AgentSpec: Speculative Decoding for Batch Inference of LLM Agents
arXiv 2608.24004coding agents Published 2026-08-25
The Empire, Long Divided, Must Unite: Architectural Convergence in Three LLM Agent Harnesses
arXiv 2608.23953quantization Published 2026-08-24
FAMPWQ: Fisher Information-based Adaptive Mixed Precision Weight Quantization for Effective LLM Inference
arXiv 2608.24945agent evaluation Published 2026-08-24
Prime Agent: A Self-Improving RLM Harness
arXiv 2608.23552memory and context Published 2026-08-24
TRACE: A Self-Evolving Skill Bank for Consistent, Limit-Aware LLM Agents
arXiv 2608.22793agent evaluation Published 2026-08-23
ToolRobustBench: Stage-Wise Perturbation Evaluation and Failure Diagnosis for Tool-Calling Agents
arXiv 2608.23635memory and context Published 2026-08-22
MemGuard: Persisting Verifier Signals for LLM-Agent Memory Governance
arXiv 2608.21867memory and context Published 2026-08-21
Context as an Environment: Programmatic Context Management for Long-Horizon Agents
arXiv 2608.21690efficient training Published 2026-08-21
Event-triggered Implicit Perturbation for Zeroth-Order Fine-Tuning of Spiking Transformers
arXiv 2608.21223speculative decoding Published 2026-08-20
LiLiCorr: Lightweight Likelihood Correlation of Parallel Drafts for Speculative Decoding
arXiv 2608.20530attention and kernels Published 2026-08-20
FlashPrefill V2: Block-Sparse Prefill Attention for Long-Context LLM Serving
arXiv 2608.19758attention and kernels Published 2026-08-19
MonoMoE: An Efficient Fused Mega-kernel for Quantized MoE Decoding
arXiv 2609.04244agent evaluation Published 2026-08-19
What is Missing from AI Post-Training AI: An Empirical Analysis
arXiv 2608.19072quantization Published 2026-08-18
Efficient INT8 Inference of Small NLP Models on Server CPUs with PyTorch Native Stack
arXiv 2608.18182memory and context Published 2026-08-15
Harness the Memory: A Holistic Evaluation of Memory Substrates in Memory Agents
arXiv 2608.15008speculative decoding Published 2026-08-14
From Positionwise Confidence to Prefix Scheduling: Verifier Skipping in Speculative Decoding
arXiv 2608.14787sparsity and pruning Published 2026-08-11
MoE Proxy Models for Low-Cost Failure Reproduction and Diagnosis in LLM RL Post-Training
arXiv 2608.10823sparsity and pruning Published 2026-08-10
UnionSparse: An Index-Efficient Sparsity Framework for Low-Bit Sparse LLM Inference on Edge
arXiv 2608.09291KV cache Published 2026-08-08
OasisKV: Scaling In-Decode KV Cache Beyond HBM with Lookahead Sparse Prefetching
arXiv 2608.08097KV cache Published 2026-08-08
CommitKV: Lifecycle-Aware KV Cache Compression via Commit Transitions for Multi-Turn Agents
arXiv 2608.07855agent evaluation Published 2026-08-07
Long-Horizon Agent Trajectory Attribution: A Unified Benchmark and Fine-Grained Annotation Framework
arXiv 2608.06909planning and decomposition Published 2026-08-04
OneDayAgent: Towards a Long-Horizon Harness for Autonomous Agents
arXiv 2608.05013retrieval and RAG Published 2026-08-04
Multimodal Alignment Through Joint Kernel Entropic Gromov--Wasserstein Optimal Transport
arXiv 2608.04234speculative decoding Published 2026-08-03
Bole: Efficient Tree Speculation for Hybrid-Attention Language Models
arXiv 2608.01651efficient training Published 2026-08-02
SAFE-Merge: Data-Free Continual Model Merging with General Knowledge Preservation
arXiv 2608.01184coding agents Published 2026-08-01
Turning Interaction History into Execution State: A Runtime Layer for Long-Horizon Coding Agents
arXiv 2608.00808speculative decoding Published 2026-07-30
A Sparse Glimpse of the Whole: Train-Free Self-Speculative Decoding
arXiv 2607.27735attention and kernels Published 2026-07-30
Recall Before You Rank: Similarity-Guided Top-$K$ Reuse for Efficient Long-Context Attention
arXiv 2607.27692speculative decoding Published 2026-07-29
Beyond KV Reconstruction: Functional Reconstruction for MLA Draft Models in Speculative Decoding
arXiv 2607.27269KV cache Published 2026-07-29
DualDecoder: Accelerate Long Context LLM Inference by Predictive Prefetch
arXiv 2607.26475speculative decoding Published 2026-07-28
AngelSpec: Towards Real-World High Performance Inference with Speculative Decoding
arXiv 2607.25852attention and kernels Published 2026-07-28
CoSA: Accelerating Long-Context Inference via Proxy-Kernel Co-Designed Sparse Attention
arXiv 2607.25291attention and kernels Published 2026-07-27
PIVOT: Efficient Query-Group Indexing for Token-Level Sparse Attention
arXiv 2607.24593KV cache Published 2026-07-24
HiKV: Hierarchical Importance-Aware KV Cache with Hardware Acceleration for LLM Decoding
arXiv 2607.22389efficient architectures Published 2026-07-14
Lightweight Multi-Scale Anomaly Detection for Resource-Constrained Edge Devices
arXiv 2607.12599speculative decoding Published 2026-06-29
When Is a Draft Accepted? A Theory of Acceptance in Speculative Decoding
arXiv 2606.30265other Published 2026-06-12
Can Deep Neural Networks Improve Compression of Very Large Scientific Data?
arXiv 2606.14353efficient architectures Published 2026-06-01
FLARE: Diffusion for Hybrid Language Model
arXiv 2606.01774other Published 2026-05-13
Di-BiLPS: Denoising induced Bidirectional Latent-PDE-Solver under Sparse Observations
arXiv 2605.13790other Published 2026-04-25
Maximizing Memory-Level Parallelism via Integrated Stochastic Logic-in-Memory Architectures
arXiv 2604.23146distillation Published 2026-03-12
Gaussian Process Regression-based Knowledge Distillation Framework for Simultaneous Prediction of Physical and Mechanical Properties of Epoxy Polymers
arXiv 2603.16925agent evaluation Published 2026-01-17
The Language You Ask In: Language-Conditioned Ideological Divergence in LLM Analysis of Contested Political Documents
arXiv 2601.12164KV cache Published 2025-08-30
KVComp: A High-Performance, LLM-Aware, Lossy Compression Framework for KV Cache
arXiv 2509.00579other Published 2023-04-18
Variational Relational Point Completion Network for Robust 3D Classification
arXiv 2304.09131
No papers match. Try another title, research area or arXiv ID.
Make the discussion useful
Link to the page, equation or code you are discussing. A good first post can be as small as: What would this change in my product?
or I tried this on [model] with [baseline] and measured [result].
When sharing a test, include the model, hardware, baseline and results. Say what you measured and what remains uncertain. Keep criticism focused on the work.
Reader comments are public contributions. They do not change the evidence status of a Paperraft brief. Repository moderators can manage discussions on GitHub.