Search: “llm”

When Terminal-Agent Training Stalls: Demystifying Data Generation and Verification Challenge

arXiv:2610.02405v1 Announce Type: new Abstract: Using a frontier model like Claude Opus as a meta-agent to generate terminal tasks and verifiers for RL training is increasingly common. Yet a runnable

5 Oct 2026

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Model Releases

A Citation-Grounded Benchmark for Trustworthy Earnings Call Transcript Analysis with Large Language Models

DGX agent

arXiv:2610.00969v1 Announce Type: cross Abstract: Large language models (LLMs) have been increasingly used for financial document analysis, including earnings call transcripts (ECTs). Beyond generatin

model-releasesarxiv-cs-ai

2 Oct 2026

Applications

A Hybrid Approach to Malware Detection: Integrating Few-Shot Model-Agnostic Meta-Learning with Autoencoders

arXiv:2610.01949v1 Announce Type: cross Abstract: Ransomware has emerged as a major cybersecurity threat, with incidents increasing in frequency and impact across critical sectors. These attacks are t

applicationsarxiv-cs-ai

A Safe Prototype Is Not a Safety Direction: Reference Dependence and Prompt Confounds in Response-Safety Embeddings

arXiv:2610.01801v1 Announce Type: new Abstract: Can response safety be scored by cosine similarity to the mean embedding of known-safe responses? A recent sleeper-agent detector proposes exactly this

safetyarxiv-cs-lg

A Simple Doxastic Deontic Logic for Norm-Guided Decision Making

arXiv:2610.00668v1 Announce Type: new Abstract: Making decisions despite conflicting norms and incomplete or unreliable information is a fundamental challenge for autonomous systems. We introduce a si

A Verifier Can Leak the Answer: Diagnosability Before Optimization in Closed-Loop Agent Debugging

arXiv:2610.00126v1 Announce Type: cross Abstract: Agent developers increasingly compare prompts, tools, policies, and diagnosis algorithms through simulator-grounded verifiers. A verifier can neverthe

Actions with Receipts: Jointly Binding Claims, Evidence, and Execution for Replayable Tool-Agent Auditing

arXiv:2610.00327v1 Announce Type: cross Abstract: Tool-using agents can expose citations and execution logs while leaving a critical association unaudited: whether the claim shown to a user is the cla

Align Then Reason: A Multimodal Lip-Sync Judge for Dubbing

model-releasesarxiv-cs-cv

arXiv:2610.00825v1 Announce Type: new Abstract: Dubbing quality control requires a reference-free judge that can determine whether a candidate text line matches a speaker's visible articulation in bot

Ask a Language Model for Lottery Numbers: Concentration in Repeated Six-of-49 Outputs

arXiv:2610.00052v1 Announce Type: cross Abstract: We evaluate six language-model configurations on requests for six distinct random integers from 1-49. Across 1,200 attempted calls using four English

Attention Manifolds: Steering or Blocking Language Models by Editing Learned B-Spline Surfaces

arXiv:2610.00257v1 Announce Type: new Abstract: In standard transformer attention, a source token sends the same value vector to every receiver. The query determines how much to attend but not what to

Benchmarking Prompt Optimization of Large Language Models With Chess

arXiv:2610.00416v1 Announce Type: new Abstract: Evaluating large language models becomes increasingly challenging as their capabilities advance: benchmarks can saturate, public test sets risk contamin

Can AI Oversight Be Zero Knowledge?

arXiv:2610.01995v1 Announce Type: new Abstract: AI systems increasingly produce outputs from confidential data, such as a fitness-for-duty assessment from medical records or the predicted properties o

Capabilities Ain't All You Need: Measuring Propensities in AI

arXiv:2602.18182v5 Announce Type: replace-cross Abstract: AI evaluation has primarily focused on measuring capabilities, with formal approaches inspired from Item Response Theory (IRT) being increasin

CAVE-Mem: Boundary-Aware Experience Validation for Memory Search

arXiv:2610.00238v1 Announce Type: cross Abstract: Long-term memory agents increasingly rely on it- erative search and reusable experience to answer questions over large personal, factual, or narrative

Comedic Fool's Gold: Reward Exploits and Countermeasures in Conversational Humor

arXiv:2610.00197v1 Announce Type: new Abstract: We investigate automated rewards for training language models in conversational humor, focusing on reward exploits and countermeasures. Two approaches a

Controllable Multi-label Video Safety Detection via Adaptive Tversky Policy Optimization

arXiv:2610.02019v1 Announce Type: cross Abstract: The rapid growth of video-based social media has increased users' exposure to harmful content, creating a need for reliable automated video safety det

safetyarxiv-cs-cv

Deny Without Disabling: Authorization-Paired Evaluation and Control for Multi-Agent Systems

arXiv:2610.00371v1 Announce Type: cross Abstract: Multi-agent systems derive their capabilities from sharing evidence, delegating tasks, and combining information across agents. The same process creat

safetyarxiv-cs-ai

Does Scaling Reinforcement Learning Really Require More Training?

arXiv:2610.01133v1 Announce Type: cross Abstract: Scaling reasoning typically spends more compute on reinforcement learning (RL) or on inference. We show that a completed RL training history can yield

Empty Commitments: When Agents Promise What Their Runtime Cannot Deliver

arXiv:2610.01045v1 Announce Type: new Abstract: A chatbot that says 'I will remind you tomorrow' will not run again until the user writes. We call such a promise an empty commitment: a promise of an a

Explainability of Complex AI Models with Correlation Impact Ratio

arXiv:2601.06701v2 Announce Type: replace-cross Abstract: Complex AI systems make better predictions but often lack transparency, limiting trustworthiness, interpretability, and safe deployment. Commo

Faynt: Scaling and Optimizing Policies for Competitive Melee

arXiv:2610.02144v1 Announce Type: new Abstract: We introduce Faynt, a family of 10M- and 75M-parameter Transformer policies for Super Smash Bros. Melee, each controlling all 26 characters with a singl

Local Ai

FedSAP: Federated Learning with Structured Adaptive Partitioning for Multi-Domain Heterogeneous Edge Devices

arXiv:2610.01638v1 Announce Type: new Abstract: Federated learning (FL) on heterogeneous edge devices must jointly accommodate unequal resource budgets and domain-shifted local data. Existing resource

local-aiarxiv-cs-lg

Local Ai

Fewer Tokens, Better Action: GPT-6 Astra Robot Agents with 14% Higher Success Rate but 65% Fewer Tokens

arXiv:2610.01939v1 Announce Type: new Abstract: Vision language model (VLM) agents can control robots through visual feedback and action primitives, but repeated model invocations and redundant observ

local-aiarxiv-cs-cv

Finding the Right Fit: Model-Harness Interactions across Agent Tasks

arXiv:2610.00917v1 Announce Type: new Abstract: Choosing an agent system means choosing both a language model and the harness through which it acts. We ask whether a strong model, harness, or pairing

Finetuning with Sampling: SFT Learns Better Than You Think

arXiv:2610.02140v1 Announce Type: cross Abstract: Introducing new capabilities to frontier models has long been the goal of posttraining, which predominantly employs supervised finetuning (SFT) and re

safetyarxiv-cs-ai

Function-Structured Reinforcement Learning with Executable Verifiers for Mathematical Reasoning

arXiv:2610.01729v1 Announce Type: new Abstract: Algorithmic mathematical reasoning requires reliable decomposition, computation, and aggregation. Final-answer rewards provide limited guidance on inter

GUI-HARVEST: Self-Improving GUI Agents through Evidence-Driven Harness Evolution

arXiv:2610.00948v1 Announce Type: cross Abstract: The executable harness surrounding a GUI model determines how observations are assembled, actions are executed, and verification, recovery, and termin

Harnessing Domain Specialists in Multimodal Mixture-of-Experts for Efficient Adaptation

arXiv:2610.02123v1 Announce Type: new Abstract: Mixture-of-Experts (MoE) architectures scale model capacity through sparse computation, routing each token through only a small subset of experts. In th

researcharxiv-cs-cv

InterviewSim: A Scalable Framework for Interview-Grounded Personality Simulation

arXiv:2602.20294v2 Announce Type: replace-cross Abstract: Simulating real personalities with large language models requires grounding generation in authentic personal data. Existing evaluation approac

safetyarxiv-cs-ai

Applications

Judgement in the Age of Jev: From Evaluation Scarcity to Evaluation Abundance

arXiv:2610.01231v1 Announce Type: cross Abstract: Generative artificial intelligence has reduced the cost of producing plausible symbolic artefacts, leading recent organisation scholarship to identify

applicationsarxiv-cs-ai

JusticeAxis: Benchmarking Legal Judgment between Rigid Rule Application and Ungrounded Discretion

arXiv:2610.00353v1 Announce Type: new Abstract: A sound judgment applies the law to established facts and weighs the circumstances in which they arose. However, existing methods swing between rigid st

Kepler: Auditable World Models for ARC-AGI-3

arXiv:2610.00834v1 Announce Type: new Abstract: ARC-AGI-3 evaluates agents in interactive environments whose rules and objectives must be inferred from observation. We present Kepler, an open-source h

Learning Rate Transfer for Hybrid Transformer-SSM Architectures

arXiv:2610.01172v1 Announce Type: new Abstract: We study learning rate (LR) scaling for hybrid architectures combining Transformer and State-Space Model (SSM) blocks, a class adopted by several recent

Applications

LineupRL: Verifiable Reinforcement Learning for Time Series Captioning via Caption-to-Series Identification

arXiv:2610.01800v1 Announce Type: new Abstract: Time series captioning is a fundamental step in time series understanding and can also serve as the bridge between signal and natural language. Supervis

applicationsarxiv-cs-ai

Measuring the Stability Assumption Behind Action Chunking

arXiv:2610.01626v1 Announce Type: new Abstract: Action chunking improves the performance of policies learned by behavioural cloning, and several mechanisms have been proposed to explain why, including

Metacognitive Reasoning in Energy Based Models using Instance Based Learning Theory

arXiv:2610.00399v1 Announce Type: new Abstract: Metacognition involves reasoning about cognitive processes themselves. An example is in resource allocation where we choose how much time and effort to

MiLoop: Selective Memory Propagation for Neural Combinatorial Optimization

arXiv:2610.01685v1 Announce Type: new Abstract: Constructive neural combinatorial optimization (NCO) has emerged as a promising paradigm that learns to construct solutions to combinatorial optimizatio

safetyarxiv-cs-lg

Mingbird: A Local-First Agent Harness Enabling Small Open Models to Complete Real Tasks

arXiv:2610.02001v1 Announce Type: new Abstract: Small open-weight models (2-9B) run on ordinary laptops, but under cloud-scale agent harnesses they rarely complete real tasks: tool prefill overflows t

Mixture-Trained Merging for Unified Multi-Objective Models

arXiv:2610.01238v1 Announce Type: new Abstract: Unified language models are increasingly expected to combine heterogeneous capabilities, such as mathematics, code, instruction following, and controlla

MOVE: Multimodal Open-world Verification and Expansion for Graph Learning

arXiv:2610.00268v1 Announce Type: new Abstract: Multimodal graph learning faces a fundamental challenge: new classes may emerge after deployment, while models are trained with a fixed label space. Exi

researcharxiv-cs-lg

Multi-Party Backchannel Prediction: a Diagnosis, a Benchmark, and a Ceiling

arXiv:2610.01488v1 Announce Type: cross Abstract: Backchannel prediction has been studied almost entirely in dyadic conversation. We introduce a multi-party benchmark based on the AMI corpus, comprisi

Not All Experience Belongs in the Weights: Component Routing for Self-Improving GUI Agents

arXiv:2610.01787v1 Announce Type: new Abstract: Self-improving GUI agents keep the trajectories they produce and return them to the agent, by fine-tuning or by retrieval into the prompt, and studies t

One pool, many targets: a conservation layer and what archival data can identify

arXiv:2610.00445v1 Announce Type: new Abstract: Pairwise guide--transcript scores do not enforce conservation of a finite guide-loaded RISC pool when they are interpreted independently as occupancies.

Permutation-Robust Decision Modeling with Candidate-Independent Block-Causal Attention

arXiv:2610.01601v1 Announce Type: cross Abstract: Decision models often score a variable-sized set of candidate actions encoded in a single sequence. This setting is increasingly relevant for System 1

Predictive Credit: Measuring What Scientific Explanations Add to Experimental Forecasts

arXiv:2610.00314v1 Announce Type: new Abstract: Research agents explain planned experiments. We measure predictive credit with paired forecasts sharing an intervention, forecaster, and outcome while v

QK-Wanda: Coupling Queries and Keys for Unstructured Pruning

arXiv:2610.01554v1 Announce Type: new Abstract: Wanda (Sun et al., 2024) prunes large language models by scoring weights independently within each linear projection, although queries and keys interact

Reducing Cognitive Overhead in Tool Use via Multi-Small-Agent Reinforcement Learning

arXiv:2508.08882v5 Announce Type: replace Abstract: Recent advances in multi-agent systems highlight the potential of specialized small agents that collaborate via division of labor. Existing tool-int

ReHoPER: Receding-Horizon Planning for Enhanced Reasoning

arXiv:2610.00940v1 Announce Type: cross Abstract: We propose ReHoPER, an inference-only, zero-shot method that improves large language models' reasoning by generating and answering intermediate questi