Search: “llm”

In-Situ Behavioral Evaluation for LLM Fairness, Not Standardized-Test Scores

arXiv:2605.12530v1 Announce Type: cross Abstract: LLM fairness should be evaluated through in-situ conversational behavior rather than standardized-test Q&A benchmarks. We show that the standardized-t

14 May 2026

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Quantifying LLM Safety Degradation Under Repeated Attacks Using Survival Analysis

DGX agent

arXiv:2605.12869v1 Announce Type: cross Abstract: Large language models (LLMs) are increasingly deployed in a wide range of applications, yet remain vulnerable to adversarial jailbreak attacks that ci

safetyarxiv-cs-ai

14 May 2026

Research

Self-CriTeach: LLM Self-Teaching and Self-Critiquing for Improving Robotic Planning via Automated Domain Generation

arXiv:2509.21543v3 Announce Type: replace Abstract: Large Language Models (LLMs) have shown strong promise for robotic task planning, particularly through the automatic generation of symbolic planning

researcharxiv-cs-ro

14 May 2026

Teacher-Guided Policy Optimization for LLM Distillation

arXiv:2605.13230v1 Announce Type: cross Abstract: The convergence of reinforcement learning and imitation learning has positioned Reverse KL (RKL) as a promising paradigm for on-policy LLM distillatio

14 May 2026

Adaptive Teacher Exposure for Self-Distillation in LLM Reasoning

arXiv:2605.11458v1 Announce Type: cross Abstract: On-policy self-distillation has become a strong recipe for LLM reasoning, where a privileged teacher supervises the student's own rollouts while condi

Breaking extit{Winner-Takes-All}: Cooperative Policy Optimization Improves Diverse LLM Reasoning

model-releasesarxiv-cs-lg

arXiv:2605.11461v1 Announce Type: cross Abstract: Reinforcement learning with verifiers (RLVR) has become a central paradigm for improving LLM reasoning, yet popular group-based optimization algorithm

GEAR: Granularity-Adaptive Advantage Reweighting for LLM Agents via Self-Distillation

arXiv:2605.11853v1 Announce Type: cross Abstract: Reinforcement learning has become a widely used post-training approach for LLM agents, where training commonly relies on outcome-level rewards that pr

Local Ai

Hierarchical LLM-Driven Control for HAPS-Assisted UAV Networks: Joint Optimization of Flight and Connectivity

arXiv:2605.11509v1 Announce Type: cross Abstract: Uncrewed aerial vehicles (UAVs) are increasingly deployed in complex networked environments, yet the joint optimization of multi-UAV motion control an

local-aiarxiv-cs-lg

MCPShield: Content-Aware Attack Detection for LLM Agent Tool-Call Traffic

arXiv:2605.11053v1 Announce Type: cross Abstract: The Model Context Protocol (MCP) has become a widely adopted interface for LLM agents to invoke external tools, yet learned monitoring of MCP tool-cal

agentsarxiv-cs-lg

PIVOT: Bridging Planning and Execution in LLM Agents via Trajectory Refinement

arXiv:2605.11225v1 Announce Type: cross Abstract: Large language model (LLM)-based agents frequently generate seemingly coherent plans that fail upon execution due to infeasible actions, constraint vi

agentsarxiv-cs-lg

PrivacySIM: Evaluating LLM Simulation of User Privacy Behavior

arXiv:2605.12147v1 Announce Type: cross Abstract: Large language models (LLMs) are increasingly used to simulate human behavior, but their ability to simulate individual privacy decisions is not well

applicationsarxiv-cs-lg

RACC: Representation-Aware Coverage Criteria for LLM Safety Testing

arXiv:2602.02280v2 Announce Type: replace-cross Abstract: Large Language Models (LLMs) face severe safety risks from jailbreak attacks, yet current safety testing largely relies on static datasets and

Robust LLM Unlearning Against Relearning Attacks: The Minor Components in Representations Matter

arXiv:2605.11685v1 Announce Type: new Abstract: Large language model (LLM) unlearning aims to remove specific data influences from pre-trained model without costly retraining, addressing privacy, copy

Hardware

The Illusion of Power Capping in LLM Decode: A Phase-Aware Energy Characterisation Across Attention Architectures

arXiv:2605.11999v1 Announce Type: cross Abstract: Power capping is the standard GPU energy lever in LLM serving, and it appears to work: throughput drops, power readings fall, and energy budgets are m

hardwarearxiv-cs-lg

VERDI: Single-Call Confidence Estimation for Verification-Based LLM Judges via Decomposed Inference

arXiv:2605.11334v1 Announce Type: cross Abstract: LLM-as-Judge systems are widely deployed for automated evaluation, yet practitioners lack reliable methods to know when a judge's verdict should be tr

Agent-Omit: Adaptive Context Omission for Efficient LLM Agents

arXiv:2602.04284v2 Announce Type: replace Abstract: Managing agent context (e.g., thought and observation) during multi-turn agent-environment interactions is an emerging strategy to improve agent eff

AgentRx: A Benchmark Study of LLM Agents for Multimodal Clinical Prediction Tasks

arXiv:2605.10286v1 Announce Type: new Abstract: Building effective clinical decision support systems requires the synthesis of complex heterogeneous multimodal data. Such modalities include temporal e

Research

Beyond Hard Writes and Rigid Preservation: Soft Recursive Least-Squares for Lifelong LLM Editing

arXiv:2601.15686v2 Announce Type: replace Abstract: Model editing updates a pre-trained LLM with new facts or rules without retraining while preserving unrelated behavior. In real deployment, edits ar

researcharxiv-cs-lg

BoostLLM: Boosting-inspired LLM Fine-tuning for Few-shot Tabular Classification

model-releasesarxiv-cs-lg

arXiv:2605.06117v2 Announce Type: replace Abstract: Large language models (LLMs) have recently been adapted to tabular prediction by serializing structured features into natural language, but their pe

ComplexMCP: Evaluation of LLM Agents in Dynamic, Interdependent, and Large-Scale Tool Sandbox

arXiv:2605.10787v1 Announce Type: new Abstract: Current LLM agents are proficient at calling isolated APIs but struggle with the 'last mile' of commercial software automation. In real-world scenarios,

Do Self-Evolving Agents Forget? Capability Degradation and Preservation in Lifelong LLM Agent Adaptation

arXiv:2605.09315v1 Announce Type: new Abstract: Recent advances in LLM agents enable systems that autonomously refine workflows, accumulate reusable skills, self-train their underlying models, and mai

EdgeFlowerTune: Evaluating Federated LLM Fine-Tuning Under Realistic Edge System Constraints

arXiv:2605.08636v1 Announce Type: new Abstract: Federated fine-tuning offers a promising paradigm for adapting large language models (LLMs) on edge devices by leveraging the rich, diverse, and continu

ER-Reason: A Benchmark Dataset for LLM Clinical Reasoning in the Emergency Room

arXiv:2505.22919v3 Announce Type: replace Abstract: Existing benchmarks for evaluating the clinical reasoning capabilities of large language models (LLMs) often lack a clear definition of 'clinical re

Human-Inspired Memory Architecture for LLM Agents

arXiv:2605.08538v1 Announce Type: new Abstract: Current LLM agents lack principled mechanisms for managing persistent memory across long interaction horizons. We present a biologically-grounded memory

LLM Agents Already Know When to Call Tools -- Even Without Reasoning

arXiv:2605.09252v1 Announce Type: new Abstract: Tool-augmented LLM agents tend to call tools indiscriminately, even when the model can answer directly. Each unnecessary call wastes API fees and latenc

PRISM: Fast Online LLM Serving via Scheduling-Memory Co-design

arXiv:2605.08581v1 Announce Type: new Abstract: Modern online large language model (LLM) services, such as Retrieval-Augmented Generation (RAG) and agent systems, increasingly expose two prominent cha

agentsarxiv-cs-lg

Local Ai

PYTHALAB-MERA: Validation-Grounded Memory, Retrieval, and Acceptance Control for Frozen-LLM Coding Agents

arXiv:2605.08468v1 Announce Type: cross Abstract: Local LLM-based coding agents increasingly work in settings where correctness is earned through execution feedback, persistent state, and bounded repa

local-aiarxiv-cs-ai

Reasoning Is Not Free: Robust Adaptive Cost-Efficient Routing for LLM-as-a-Judge

arXiv:2605.10805v1 Announce Type: new Abstract: Reasoning-capable large language models (LLMs) have recently been adopted as automated judges, but their benefits and costs in LLM-as-a-Judge settings r

Shaping Schema via Language Representation as the Next Frontier for LLM Intelligence Expanding

arXiv:2605.09271v1 Announce Type: new Abstract: Although natural language is the default medium for Large Language Models (LLMs), its limited expressive capacity creates a profound bottleneck for comp

applicationsarxiv-cs-ai

SkillLens: Adaptive Multi-Granularity Skill Reuse for Cost-Efficient LLM Agents

arXiv:2605.08386v1 Announce Type: new Abstract: Skill libraries have become a practical way for LLM agents to reuse procedural experience across tasks. However, existing systems typically treat skills

agentsarxiv-cs-ai

SkillMAS: Skill Co-Evolution with LLM-based Multi-Agent System

arXiv:2605.09341v1 Announce Type: cross Abstract: Large language model (LLM) agent systems are increasingly expected to improve after deployment, but existing work often decouples two adaptation targe

agentsarxiv-cs-cl

SmartEval: A Benchmark for Evaluating LLM-Generated Smart Contracts from Natural Language Specifications

arXiv:2605.09610v1 Announce Type: cross Abstract: We introduce SmartEval, a benchmark for systematically evaluating the quality of Solidity smart contracts generated by large language models (LLMs) fr

SPECTRE: Hybrid Ordinary-Parallel Speculative Serving for Resource-Efficient LLM Inference

arXiv:2605.08151v1 Announce Type: cross Abstract: LLM serving platforms are increasingly deployed as multi-model cloud systems, where user demand is often long-tailed: a few popular large models recei

applicationsarxiv-cs-ai

The Art of the Jailbreak: Formulating Jailbreak Attacks for LLM Security Beyond Binary Scoring

arXiv:2605.09225v1 Announce Type: cross Abstract: Jailbreak attacks -- adversarial prompts that bypass LLM alignment through purely linguistic manipulation -- pose a growing operational security threa

TodyComm: Task-Oriented Dynamic Communication for Multi-Round LLM-based Multi-Agent System

arXiv:2602.03688v2 Announce Type: replace Abstract: Multi-round LLM-based multi-agent systems rely on effective communication structures to support collaboration across rounds. However, most existing

Token Economics for LLM Agents: A Dual-View Study from Computing and Economics

arXiv:2605.09104v1 Announce Type: new Abstract: As LLM agents evolve, tokens have emerged as the core economic primitives of Agentic AI. However, their exponential consumption introduces severe comput

agentsarxiv-cs-ai

Hardware

Understand and Accelerate Memory Processing Pipeline for Disaggregated LLM Inference

arXiv:2603.29002v2 Announce Type: replace-cross Abstract: Modern large language models (LLMs) increasingly depends on efficient long-context processing and generation mechanisms, including sparse atte

hardwarearxiv-cs-ai

Why Retrying Fails: Context Contamination in LLM Agent Pipelines

arXiv:2605.08563v1 Announce Type: new Abstract: When an LLM agent fails a multi-step tool-augmented task and retries, the failed attempt typically remains in its context window -- contaminating the ne

Your Simulation Runs but Solves the Wrong Physics: PDE-Grounded Intent Verification for LLM-Generated Multiphysics Simulation Code

arXiv:2605.09360v1 Announce Type: cross Abstract: Execution-based evaluation of LLM-generated code implicitly treats successful execution as a proxy for correctness. In scientific simulation, this pro

Active Learning for Communication Structure Optimization in LLM-Based Multi-Agent Systems

arXiv:2605.05703v2 Announce Type: replace-cross Abstract: Optimizing the communication structure of large language model based multi-agent systems (LLM-MAS) has been shown to improve downstream perfor

agentsarxiv-cs-ai

Bias and Uncertainty in LLM-as-a-Judge Estimation

arXiv:2605.06939v1 Announce Type: new Abstract: LLM-as-a-Judge evaluation has become a standard tool for assessing base model performance. However, characterizing performance via the naive estimator,

safetyarxiv-cs-lg

Hardware

Dooly: Configuration-Agnostic, Redundancy-Aware Profiling for LLM Inference Simulation

arXiv:2605.07985v1 Announce Type: cross Abstract: Selecting the optimal LLM inference configuration requires evaluation across hardware, serving engines, attention backends, and model architectures, s

hardwarearxiv-cs-ai

EnvSimBench: A Benchmark for Evaluating and Improving LLM-Based Environment Simulation

arXiv:2605.07247v1 Announce Type: new Abstract: Scalable AI agents training relies on interactive environments that faithfully simulate the consequences of agent actions. Manually crafted environments

Evaluating Prompt Injection Defenses for Educational LLM Tutors: Security-Usability-Latency Trade-offs

arXiv:2605.06669v1 Announce Type: cross Abstract: Educational LLM tutors face a core AI alignment challenge: they must follow user intent while preserving pedagogical constraints and safety policies.

EvolveR: Self-Evolving LLM Agents through an Experience-Driven Lifecycle

arXiv:2510.16079v2 Announce Type: replace-cross Abstract: Current Large Language Model (LLM) agents show strong performance in tool use, but lack the crucial capability to systematically learn from th

LLM-Based Agents for Competitive Landscape Mapping in Drug Asset Due Diligence

arXiv:2508.16571v4 Announce Type: replace Abstract: In this paper, we describe and benchmark a competitor-discovery component used within an agentic AI system for fast drug asset due diligence. A comp

LLM hallucinations in the wild: Large-scale evidence from non-existent citations

arXiv:2605.07723v1 Announce Type: cross Abstract: Large language models (LLMs) are known to generate plausible but false information across a wide range of contexts, yet the real-world magnitude and c

applicationsarxiv-cs-ai

Mage: Multi-Axis Evaluation of LLM-Generated Executable Game Scenes Beyond Compile-Pass Rate

arXiv:2605.07342v1 Announce Type: cross Abstract: Compile-pass rate is the dominant evaluation signal for LLM code generation, yet for multi-component domain-specific artifacts it can be actively misl