Search: “llm”

arXiv:2603.09221v2 Announce Type: replace Abstract: Associative memory has long underpinned the design of sequential models. Beyond recall, humans reason by projecting future states and selecting goal

hardwarearxiv-cs-lg

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Consolidating Rewarded Perturbations for LLM Post-Training

DGX agent

arXiv:2605.31494v1 Announce Type: new Abstract: Post-training of language models is commonly framed as a sample-score-update loop implemented by gradient descent. A recent line of work, exemplified by

researcharxiv-cs-cl

1 Jun 2026

Counterfactual Trace Auditing of LLM Agent Skills

arXiv:2605.11946v2 Announce Type: replace Abstract: Large Language Model agents are increasingly augmented with agent skills. Current evaluation methods for skills remain limited. Most deployed benchm

Applications

D^3: Dynamic Directional Graph-Constrained Data Scheduling for LLM Training

arXiv:2605.31164v1 Announce Type: cross Abstract: Training data plays a central role in large language models (LLMs) optimization, motivating extensive research on data scheduling strategies. Most exi

applicationsarxiv-cs-ai

DARTS: Distribution-Aware Active Rollout Trajectory Shaping for Accelerating LLM Reinforcement Learning

arXiv:2605.30859v1 Announce Type: cross Abstract: Reinforcement Learning (RL) has become pivotal for improving model capabilities yet suffers from rollout efficiency bottlenecks due to the long-tail r

researcharxiv-cs-ai

Distilling LLM Feedback for Lean Theorem Proving

arXiv:2605.30861v1 Announce Type: new Abstract: Post-training for reasoning models typically combines supervised fine-tuning with reinforcement learning from verifiable rewards, most commonly with GRP

LLM Judges Inconsistently Disagree Across Safety Criteria and Harm Categories

arXiv:2605.31381v1 Announce Type: new Abstract: We evaluate the consistency of automated judges in conducting a multi-dimensional safety evaluation in a reference-free setup. Our results indicate that

Local Ai

Memory-Efficient Structured Backpropagation for On-Device LLM Fine-Tuning

arXiv:2602.13069v2 Announce Type: replace-cross Abstract: On-device fine-tuning enables privacy-preserving personalization of large language models, but mobile devices impose severe memory constraints

local-aiarxiv-cs-cl

OBCache: Optimal Brain KV Cache Pruning for Efficient Long-Context LLM Inference

arXiv:2510.07651v2 Announce Type: replace-cross Abstract: Large language models (LLMs) with extended context windows enable powerful applications but impose significant memory overhead, as caching all

Re-examining Low Rank adaptation for private LLM fine-tuning

model-releasesarxiv-cs-lg

arXiv:2510.01137v3 Announce Type: replace Abstract: Privacy is a central concern when fine-tuning large language models (LLMs) on sensitive data, and differentially private stochastic gradient descent

TSM-Bench: Detecting LLM-Generated Text in Real-World Wikipedia Editing Practices

arXiv:2605.31113v1 Announce Type: new Abstract: Automatically detecting machine-generated text (MGT) is critical to maintaining the knowledge integrity of user-generated content (UGC) platforms such a

Aligned but Fragile: Enhancing LLM Safety Robustness via Zeroth-Order Optimization

arXiv:2605.29396v1 Announce Type: new Abstract: Safety alignment for large language models (LLMs) aims to reduce harmful or unsafe behavior while preserving general utility. However, recent findings r

Analyzing Persona Effects in Generated Explanations from Multimodal LLM Agents in Urban Perception

arXiv:2605.29064v1 Announce Type: new Abstract: We study how persona prompting shapes language generated by multimodal large language models in an urban perception setting. Using 59,808 annotations fr

researcharxiv-cs-cl

Anytime-Valid Federated Conformal RAG for LLM Swarms

arXiv:2605.29139v1 Announce Type: cross Abstract: Federated Conformal RAG (FC-RAG) provides distribution-free coverage for a bandwidth-limited swarm of weak language models, but only at a fixed horizo

safetyarxiv-cs-lg

AttuneBench: A Conversation-Based Benchmark for LLM Emotional Intelligence

arXiv:2605.21739v2 Announce Type: replace Abstract: Emotional intelligence (EI), the ability to perceive, understand, and respond appropriately to others' emotional states, is central to human communi

Benchmarking LLM-Assisted Blue Teaming via Standardized Threat Hunting

arXiv:2509.23571v3 Announce Type: replace-cross Abstract: As cyber threats continue to grow in scale and sophistication, blue team defenders increasingly require advanced tools to proactively detect a

Beyond Attack Success Rate: Temporal Logit Observability for LLM Safety Failures

arXiv:2605.29629v1 Announce Type: new Abstract: Attack Success Rate (ASR) evaluates each jailbreak with a single yes/no label at the end of generation, telling us whether a failure happened but not ho

FinGuard: Detecting Financial Regulatory Non-Compliance in LLM Interactions

arXiv:2605.29427v1 Announce Type: new Abstract: As large language models (LLMs) are increasingly deployed in financial services, a single non-compliant interaction can expose institutions to regulator

Frontier LLM-based agents can overcome the ontology curation bottleneck for natural phenotypes

arXiv:2605.28965v1 Announce Type: new Abstract: Linking free-text phenotype descriptions to ontology terms, typically referred to as phenotype annotation, is essential for the cross-study integration

GRUFF: LLM Pronoun Fidelity, Reasoning, and Biases in German

arXiv:2605.30214v1 Announce Type: new Abstract: Third-person singular pronouns have long been used to study stereotypical biases in language models and to test their abilities to reason about referenc

HARP: Hadamard-Preconditioned Adaptive Rotation Processor for Extreme LLM Quantization

arXiv:2605.29843v1 Announce Type: cross Abstract: Post-training quantization (PTQ) is essential for deploying LLMs under memory and bandwidth constraints. However, extreme low-bit quantization remains

researcharxiv-cs-ai

How LoRA Remembers? A Parametric Memory Law for LLM Finetuning

arXiv:2605.30260v1 Announce Type: cross Abstract: Large Language Models (LLMs) must continuously learn and update knowledge to remain effective in dynamic real-world environments. While Low-Rank Adapt

InsightEval: An Expert-Curated Benchmark for Assessing Insight Discovery in LLM-Driven Data Agents

arXiv:2511.22884v2 Announce Type: replace Abstract: Data analysis has become an indispensable part of scientific research. To discover the latent knowledge and insights hidden within massive datasets,

Internal Representation, Not Clinical Knowledge: Where Apparent LLM Triage Failures Originate

arXiv:2605.29889v1 Announce Type: cross Abstract: Patient-voiced clinical-triage benchmarks report high under-triage rates for consumer LLMs for constrained multiple-choice output, yet the same cases

Is Your LLM Overcharging You? Tokenization, Transparency, and Incentives

arXiv:2505.21627v4 Announce Type: replace-cross Abstract: State-of-the-art large language models require specialized hardware and substantial energy to operate. As a consequence, cloud-based services

Tutorials

Knowing What to Solve Before How: Preplan Empowered LLM Mathematical Reasoning

arXiv:2605.30245v1 Announce Type: new Abstract: Current plan-based reasoning methods improve large language models (LLMs) by inserting a planning stage before execution, giving rise to the question ri

tutorialsarxiv-cs-cl

Less is Enough: Synthesizing Diverse Data in LLM Feature Space with Sparse Autoencoders

arXiv:2602.10388v3 Announce Type: replace-cross Abstract: The diversity of post-training data is critical for effective downstream performance in large language models (LLMs). Many existing approaches

Lightweight Multimodal LLM-Enabled Cost-Effective Defect Grading of Power Transmission Equipment

arXiv:2605.28822v1 Announce Type: new Abstract: Defect grading of power transmission equipment (DGPTE) is crucial to the stability of electric energy transmission. Although existing machine learning m

researcharxiv-cs-cl

LLUMI: Improving LLM Writing Assistance for Mental Health Support with Online Community Feedback

arXiv:2605.30273v1 Announce Type: cross Abstract: Large language models (LLMs) show promise in generating supportive responses for mental health queries, but improving their usefulness, empathy, and s

Tutorials

Make LLM Learn to Synthesize from Streaming Experiences through Feedback

arXiv:2605.29940v1 Announce Type: new Abstract: Large language models (LLMs) have been widely adopted for synthetic data generation, significantly reducing annotation costs. However, most existing stu

tutorialsarxiv-cs-ai

SciIntBench: Measuring LLM Compliance with Research Integrity Norms Under Adversarial Framing

arXiv:2605.29468v1 Announce Type: cross Abstract: Large language models (LLMs) are increasingly used to support scientific work, but it is unclear whether they uphold responsible conduct of research (

TRACE: Toulmin-based Reasoning Assessment through Constructive Elements for LLM CoT Evaluation

arXiv:2605.29656v1 Announce Type: new Abstract: Evaluating open-ended outputs from large language models (LLMs) remains challenging due to the absence of ground truth. Existing metrics rely on final-a

Agents

VikingMem: A Memory Base Management System for Stateful LLM-based Applications

arXiv:2605.29640v1 Announce Type: new Abstract: Large Language Models have revolutionized interactive applications; however, their finite context windows pose a critical data management challenge for

agentsarxiv-cs-ai

Local Ai

A Paired Testing Protocol for Batch-Conditioned Refusal Robustness in LLM Serving

arXiv:2605.27763v1 Announce Type: new Abstract: Safety evaluations of language models often treat serving configuration as fixed background infrastructure, but batch condition is an untested treatment

local-aiarxiv-cs-lg

AssertLLM2: A Comprehensive LLM Benchmark for Assertion Generation from Design Specifications

arXiv:2605.27472v1 Announce Type: cross Abstract: Assertion-based verification (ABV) is a cornerstone of modern hardware design, yet manually translating design intent into formal SystemVerilog Assert

Beyond One Path: Evaluating and Enhancing Divergent Thinking in Interactive LLM Agents

arXiv:2605.28465v1 Announce Type: new Abstract: Divergent thinking is a core dimension of creativity, yet existing evaluations of Large Language Models (LLMs) treat them as single-turn text generation

DeepSciVerify: Verifying Scientific Claim--Citation Alignment via LLM-Driven Evidence Escalation

arXiv:2605.27710v1 Announce Type: new Abstract: Misalignment between claims and their cited evidence is a common failure mode in reports generated by large language models, limiting their reliability

Agents

Defending LLM-based Multi-Agent Systems Against Cooperative Attacks with Sentence-Level Rectification

arXiv:2605.28104v1 Announce Type: new Abstract: Recent years have witnessed the rapid development of Large Language Model-based Multi-Agent Systems (MAS), which excel at collaborative decision-making

agentsarxiv-cs-ai

DisasterBench: Benchmarking LLM Planning under Typed Tool Interface Constraints

arXiv:2605.27957v1 Announce Type: new Abstract: Disasters cause severe societal impacts, demanding rapid coordination of heterogeneous AI tools, from satellite analysis to flood prediction and damage

Disentangling Language Roles in Multilingual LLM Task Execution

arXiv:2605.27649v1 Announce Type: new Abstract: Multilingual LLMs are increasingly used when instruction, source content, and required response languages do not coincide. Existing benchmarks have expa

FABSVer: Faster Training and Better Self-Verification for LLM Mathematical Reasoning

arXiv:2605.28389v1 Announce Type: new Abstract: While large language models have made significant progress in mathematical reasoning, they remain unreliable at judging the correctness of their own sol

Geometry of Human Perceptual Domains Emerges Transiently in LLM Representations

arXiv:2605.27970v1 Announce Type: new Abstract: While large language models (LLMs) are trained purely on textual data, prior work has shown that their internal representations can exhibit rich geometr

Agents

Knowing When to Ask: Segment-Level Credit Assignment for LLM Tool Use

arXiv:2605.27788v1 Announce Type: cross Abstract: Humans know when to reach for help e.g. 347 imes 28 warrants a calculator while 2+2 does not. Language models do not. Prompt-based approaches can inst

agentsarxiv-cs-cl

Learning to Translate from Soft to Hard LLM Prompts

arXiv:2605.27642v1 Announce Type: new Abstract: Soft prompt tuning is a parameter-efficient method for adapting LLMs to specific tasks, but suffers from a lack of interpretability. Building on recent

Narrative Flattening: How Post-Training Compresses Thematic, Affective, and Stylistic Variation in LLM Fiction

arXiv:2605.27878v1 Announce Type: new Abstract: Large language models produce fluent fiction, yet their creative output is widely seen as flat. We ask where this quality originates in the training and

PortBench: A Correlation-Aware, Full-Pipeline Benchmark for LLM-Driven Portfolio Management

arXiv:2605.27887v1 Announce Type: new Abstract: LLMs have shown strong performance across diverse financial tasks, yet portfolio management (PM), a critical financial decision-making task, remains poo

Prefix-Safe Bayesian Belief Tracking for LLM Reasoning Reliability:Separating Calibration from Ranking

arXiv:2605.27712v1 Announce Type: new Abstract: Long reasoning traces need reliability estimates before final answers are known. We study prefix-conditioned eventual-success estimation, P(y=1 mid o_{1

researcharxiv-cs-ai

Restoring the Sweet Spot: Pass-Rate Weighted Self-Distillation for LLM Reasoning

arXiv:2605.27765v1 Announce Type: cross Abstract: Self-Distillation Policy Optimization (SDPO) provides dense token-level credit assignment for reinforcement learning with large language models by lev