Search: “arxiv-cs-ai”

FSPO: Policy-Consistent Risk and Pareto-Feasible Control for Budgeted LLM RL Post-Training

arXiv:2610.02828v1 Announce Type: new Abstract: Adaptive LLM reinforcement-learning post-training changes multiple training actuators online, including rollout temperature, group size, clipping, KL re

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

Gains and Collapse in On-Policy Distillation:A Reinforcement Learning Perspective

DGX agent

arXiv:2610.03185v1 Announce Type: new Abstract: On-policy distillation (OPD) has become an important approach to language model post-training. However, despite its performance gains, OPD can also coll

safetyarxiv-cs-ai

5 Oct 2026

Geometry-Aware Time Reparameterization for Flow-Map Distillation

arXiv:2610.02427v1 Announce Type: cross Abstract: Flow-map distillation enables one- and few-step generation by learning finite-time transitions of a pretrained generative ODE. We investigate whether

Geometry Meets Physics: Data-Efficient Pre-Training for Unstructured Neural PDE Solvers

arXiv:2610.03363v1 Announce Type: new Abstract: Neural surrogate models for Partial Differential Equations (PDEs) on unstructured 3D geometries are often limited by poor generalization and the high co

GISTBench: Evaluating LLM User Understanding via Evidence-Based Interest Verification

arXiv:2603.29112v2 Announce Type: replace Abstract: We introduce GISTBench, a benchmark for evaluating Large Language Models' (LLMs) ability to understand users from their interaction histories in rec

hacktrace: behavior-supervised detection of reward hacking during code generation

arXiv:2610.03055v1 Announce Type: new Abstract: A coding agent can earn a passing grade by fixing its code, or by deleting the test that exposes the bug. Detecting such reward hacking requires recogni

Hardware

Hardware-Native Joint Sparse-Quantization for Trillion-Scale Mixture-of-Experts

arXiv:2610.02241v1 Announce Type: cross Abstract: Mixture-of-Experts (MoE) architectures allow frontier language models to scale to trillions of parameters, but their deployment is constrained by mass

hardwarearxiv-cs-ai

Harness-Aware Distillation for Small Language Model Agents

arXiv:2610.02858v1 Announce Type: new Abstract: Language model agents are deployed with a harness, the software around the model that manages its context, tools, and feedback. When such an agent is di

HASTE: Evolving Agent Harnesses Against Emerging Attacks Using Sparse Evidence

arXiv:2610.02920v1 Announce Type: new Abstract: Agent harnesses play a critical role in defenses by enforcing safety constraints to prevent unsafe actions. However, rapidly emerging attacks outpace ma

HazardWeaver: Scientific Route Selection for Hazard Analysis Agents

arXiv:2610.03591v1 Announce Type: new Abstract: Understanding and assessing natural hazards is essential for disaster preparedness and risk reduction. Recent advances in large language models have spu

Heads, Tails, and AI Fails: LLMs, Randomness, and Human Judgments

arXiv:2406.00092v2 Announce Type: replace Abstract: Randomness is central to human cognition and to many applications in which large language models are deployed, yet probabilistic token generation do

Hop-Decayed Influence: New Vulnerabilities of Structural Auxiliary Indexing in GraphRAG Pipelines with LLM

arXiv:2610.02373v1 Announce Type: cross Abstract: GraphRAG pipelines construct auxiliary structures during offline indexing--semantic summaries, hierarchical edges, and pre-computed scores--that deter

Local Ai

How Artificial Intelligence LLM Engines Shape the Global Conflict Information Environment

arXiv:2607.14197v2 Announce Type: replace Abstract: Artificial Intelligence (AI) answer engines now field a growing share of the questions that analysts, scholars, and the public ask about issues of p

local-aiarxiv-cs-ai

How Causality Bridges the Semantic Gap

arXiv:2610.02594v1 Announce Type: cross Abstract: Numerical measurements capture how a system behaves, but often leave the meanings of its variables unspecified. Some variables are measured but never

How to Find and Reuse Policies for Continuous Adaptation in Lifelong Reinforcement Learning

arXiv:2610.03119v1 Announce Type: cross Abstract: In lifelong reinforcement learning, retaining previously learned policies is not sufficient for effective transfer to a new task. Useful knowledge may

How to Have a Sensitive Debate: An Instance-Optimal Protocol for AI Debate

arXiv:2610.02557v1 Announce Type: new Abstract: As powerful AI systems reach and sometimes surpass the abilities of human experts across a range of cognitively demanding tasks, the problem of accurate

How To Train Your World Model: Fine-tuning vs RAG for LM-based World Modeling

arXiv:2610.02542v1 Announce Type: new Abstract: World models (WMs) simulate the transition dynamics of environments, enabling agents to plan over the consequences of their actions. In text-based envir

Local Ai

HXAI: Hierarchical Privacy-Preserving Explainable AI in Distributed Energy Systems

arXiv:2610.02504v1 Announce Type: new Abstract: Balancing electricity demand and supply is increasingly difficult due to the inherent intermittency of renewable power generation and the stochastic pow

local-aiarxiv-cs-ai

Hybrid Reasoning Systems That Prioritize and Enhance Human Intelligence

arXiv:2504.13477v3 Announce Type: replace-cross Abstract: In a world of accelerating change, there is a need for wise and adaptive human reasoning. Integrating AI capabilities with human guidance offe

HyperBrowseComp: A Multilingual and Multimodal Stress Test for Web-Browsing Agents

arXiv:2610.03574v1 Announce Type: new Abstract: We introduce HyperBrowseComp, a multilingual and multimodal browsing benchmark comprising 423 manually authored and human-validated questions across 13

Hypothesis-guided discovery of cognitive algorithms via program refinement

arXiv:2610.02523v1 Announce Type: new Abstract: Developing cognitive models of algorithmic reasoning from behavioral data is a central problem in cognitive science that challenges current methods. Tra

'I just assumed that it would translate': examining MT risk awareness among healthcare staff with abbreviations as a use case

arXiv:2610.02496v1 Announce Type: new Abstract: In the UK, public healthcare staff report turning to machine translation (MT) - predominantly Google Translate (GT) - to communicate with patients acros

IGNITE Tokamak World Model Architecture

arXiv:2610.02515v1 Announce Type: cross Abstract: We introduce IGNITE, a generative world foundation model for fusion plasma behavior simulation trained in a self-supervised manner from over a decade

Improving Atomic-Fact Recall via Focused Views in Unstructured Knowledge Editing

arXiv:2610.02772v1 Announce Type: cross Abstract: Large language models (LLMs) increasingly serve as general-purpose interfaces to factual knowledge, but their parameters do not automatically reflect

Improving the Energy-Efficiency of the Code Generated by LLMs through Effective Prompting

arXiv:2610.02571v1 Announce Type: cross Abstract: As AI-assisted programming becomes increasingly mainstream, the environmental impact of AI-generated software has emerged as an important consideratio

Information Limits of Low-Rank Approximation Certification

arXiv:2610.03321v1 Announce Type: cross Abstract: Low-rank approximation can require additional matrix--vector products to verify that its error meets a prescribed tolerance. We characterize this cert

Agents

Inherit-MAS: Test-Time Evolution of Multi-Agent Systems through Workflow and Execution Inheritance

arXiv:2610.02396v1 Announce Type: cross Abstract: Multi-agent systems (MAS) built from large language models coordinate specialized agents to tackle complex tasks, but effective workflows are difficul

agentsarxiv-cs-ai

Instance-Dependent Regret for CMDPs with Step-Wise Constraints

arXiv:2610.02520v1 Announce Type: cross Abstract: We study online learning in episodic tabular constrained Markov decision processes with step-wise safety constraints. In such a setting, the constrain

IntentCoding: Amplifying User Intent in Code Generation

arXiv:2602.00066v1 Announce Type: cross Abstract: Large Language Models (LLMs) have shown strong capabilities in code generation, but their adherence to fine-grained user intent with multiple constrai

Interpreting at Write Time: A Policy Ablation for Multi-Goal Agent Memory

arXiv:2610.02897v1 Announce Type: new Abstract: A long-running assistant cannot keep everything it has seen, so it summarises. Summarising is not neutral: what is kept is chosen against some notion of

iS-KV: Online Low-Rank KV Cache Compression via Block-Incremental SVD

arXiv:2610.02815v1 Announce Type: new Abstract: Long chain-of-thought reasoning substantially increases KV-cache memory during autoregressive decoding, as every generated token introduces new key and

JOVE: Joint Execution and Verification for Resource-Aware LLM Task Graphs

arXiv:2610.03296v1 Announce Type: new Abstract: Complex reasoning queries can be decomposed into directed acyclic task graphs and distributed across heterogeneous LLMs, reducing latency through parall

Applications

Jumping the Line: Exploiting Length Predictions in LLM Scheduling

arXiv:2610.03430v1 Announce Type: new Abstract: Efficient request scheduling is increasingly important for reducing completion time in large language model (LLM) serving. Size-based policies such as S

applicationsarxiv-cs-ai

Keep It CALM: Analyzing the Limits of Global Unsafety in Text-to-Image Generation

arXiv:2610.02300v1 Announce Type: new Abstract: Training-free safeguards for text-to-image generation often rely on a reusable safety signal, such as an unsafe direction or global toxic subspace, appl

Keeping JEPA World Models Plannable When Little of the Frame Moves

arXiv:2610.03137v1 Announce Type: new Abstract: Specifying a goal in language rather than as a goal frame is a natural interface for planning with a latent world model, but testing it needs scenes in

Agents

Knowledge or Calculator? Decomposing the Skill Premium in Verifiable Financial Agent Workflows

arXiv:2610.03564v1 Announce Type: new Abstract: Financial AI agents must do more than retrieve facts: investment workflows require correct quantitative execution, reliable use of procedural resources,

agentsarxiv-cs-ai

KV^2: A Self-Refining KV Cache

arXiv:2610.03198v1 Announce Type: new Abstract: The memory footprint of the key-value (KV) cache constrains the practical use of long-context models, and it dominates cost when one prefilled context m

Label-Efficient Time Series Classification at Scale: A Dual-Stream OSSE-LSTM with Counterfactual Attribution

arXiv:2610.02704v1 Announce Type: new Abstract: Time series are produced continuously at enormous scale by industrial equipment, wearables, power grids, and clinical monitors, yet annotation remains m

Labels Override Definitions in Jev-Style Typed Decision Models

arXiv:2610.02586v1 Announce Type: new Abstract: A typed decision model answers a fixed question about an input by returning a probability for each of several caller-defined options. Each option carrie

Large Language Continuous Diffusion Models

arXiv:2610.02665v1 Announce Type: cross Abstract: Despite the success of discrete diffusion language models (dLMs) for fast parallel decoding, their non-smooth, high-dimensional space hinders trajecto

Large language models exhibit unreliable updating of clinical judgment as patient evidence evolves

arXiv:2610.02684v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly explored for clinical reasoning, but whether they appropriately revise judgments as patient evidence evolv

Local Ai

Law And Order: Tax Law Autoformalization

arXiv:2610.02792v1 Announce Type: new Abstract: Legal systems are increasingly implemented through software, yet scalable methods for translating legal texts into accurate symbolic representations rem

local-aiarxiv-cs-ai

Agents

LEAP: Learning Efficient Action Proposals For LLM Agents

arXiv:2610.02670v1 Announce Type: cross Abstract: LLM agents are known to be slow in rollouts. An agent completes a task one step at a time. At each step, it reasons and then chooses an action to exec

agentsarxiv-cs-ai

Learning a Fact Is Not Learning How to Retrieve It

arXiv:2610.03251v1 Announce Type: new Abstract: A model trained on 'The capital of X is Y' may produce 'Y' after 'The capital of X is' but fail after 'The capital of X:'. We call these different ways

Learning Low-Frequency Motion Control for Robust and Dynamic Robot Locomotion

arXiv:2209.14887v3 Announce Type: replace-cross Abstract: Robotic locomotion is often approached with the goal of maximizing robustness and reactivity by increasing motion control frequency. We challe

Learning Style, Forgetting Semantics: A Case Study of SFT and RFT on Classification Tasks

arXiv:2610.02437v1 Announce Type: cross Abstract: Why does supervised fine-tuning (SFT) lead to more forgetting than reinforcement fine-tuning (RFT), even when all teacher demonstrations are semantica

Applications

Learning to Assess Heartbeat Observability for mmWave Heart-Rate Sensing

arXiv:2610.03570v1 Announce Type: new Abstract: Contactless heart-rate sensing with millimeter-wave (mmWave) radar requires assessing whether individual measurements support reliable estimation. We st

applicationsarxiv-cs-ai

Learning to Revise Reasoning with Segment-wise On-Policy Distillation

arXiv:2610.02703v1 Announce Type: new Abstract: On-policy distillation (OPD) improves large language model reasoning by training students on their own rollouts with dense token-wise supervision from t