Every paper checked — nothing hidden
All papers
Every paper checked for brief #4 of 12–19 Oct 2026 — nothing is hidden. Brief papers first.
All17,312In brief24More (on-topic)3,333Not relevant13,753Not assessed87Older351Could not confirm93Blogs with no paper266All blogs284
| Paper | Topic | Relevance | Feedback |
|---|---|---|---|
| Causal Syntax Circuits in CodeLMs: Budget and Probe Decide What Looks Localised This paper investigates internal model circuits and prompt properties that drive code language models to default to Python code output. Ranked #117Surprising resultPreprint5/10 suggested | LLMs for code via arxiv · 17 Oct, 06:18 | On-topic Analyzes internal circuit mechanisms underlying programming language preferences in code LLMs. | |
| Beyond Reachability: A Relation-Semantic Reasoning Framework for Biomedical Multi-hop QA The paper proposes a relation-semantic reasoning framework to resolve disconnects between answer reachability and valid paths in biomedical graphs. Ranked #117Preprint0/10 suggested | Retrieval-augmented generation (RAG) via arxiv · 15 Oct, 15:06 | On-topic Addresses multi-hop retrieval and evidence reasoning over knowledge graphs for question answering. | |
| Full-duplex Spoken Language Modeling with Discrete Diffusion The paper presents Diffuplex, a full-duplex spoken language model that separates speech timing from content generation using discrete diffusion. Ranked #117New way to see a problemPreprint6/10 suggested | Small and efficient models via arxiv · 15 Oct, 06:17 | On-topic Designs discrete diffusion decoding that parallelizes speech generation to achieve high token throughput and low latency. | |
| When Pleasing the User Compromises Care: AI Sycophancy in Caregiver-Mediated Healthcare The paper benchmarks how caregiver-directed language models compromise medical advice when users express clinically inferior care preferences. Ranked #117Surprising resultPreprint7/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:17 | On-topic Investigates safety and alignment failures where LLMs compromise user care to satisfy harmful user requests. | |
| CaptchaAgent: Verifiable Visual Reasoning for CAPTCHA Solving Presents a structured visual reasoning agent that uses intermediate symbolic representations to solve complex visual CAPTCHAs. Ranked #117Surprising resultPreprint6/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Addresses structured visual reasoning and grounding in vision-language model agents. | |
| Accurate but Not Humble: Evaluating Epistemic Humility in LLM Agents under Knowledge Conflict Proposes an evaluation framework to measure how LLM agents handle uncertainty and resolve conflicting knowledge during execution. Ranked #117New way to see a problemPeer-reviewed7/10 suggested | LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28 | On-topic Introduces a new evaluation framework and behavioral dimensions to benchmark agent uncertainty. | |
| X-AgentRed: Red-Teaming Personal Agents in Evolving Multi-Service Ecosystems Presents an automated red-teaming framework that searches for harmful compositional tool-use paths across multi-service environments. Ranked #117New testPreprint7/10 suggested | AI agents via arxiv · 12 Oct, 16:28 | On-topic Addresses multi-service tool interactions and unintended compositional actions in personal agents. | |
| XViewRank: Overlap-Conditioned Score Fusion for LLM-Augmented Sparse Retrieval XViewRank uses result-set overlap context to fuse BM25 rankings generated from distinct query perspectives created by language models. Ranked #118Preprint0/10 suggested | Retrieval-augmented generation (RAG) via arxiv · 15 Oct, 15:06 | On-topic Improves sparse retrieval ranking using LLM-generated query views for knowledge retrieval. | |
| Intent-Triggered Anchoring for Multi-Token Format Units in Parallel Decoding of Diffusion Language Models This paper introduces a training-free decoding rule that anchors multi-token format units to ensure parseable parallel generation in diffusion language models. Ranked #118Surprising resultPreprint6/10 suggested | Small and efficient models via arxiv · 15 Oct, 06:18 | On-topic Improves parallel decoding in diffusion language models to maintain structured output formatting efficiently. | |
| Natural Code Transformation Confuses Agents in Vulnerability Analysis The study reports that natural code transformations confuse vulnerability analysis agents, proposing functional changes as a potential defensive strategy. Ranked #118Surprising resultPreprint7/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:17 | On-topic Examines defenses against adversarial agentic attacks on software vulnerability discovery. | |
| REINS: Agent-guided Thread Scheduling for Concurrency Bug Detection and Reproduction Uses agentic reasoning over program concurrency semantics to steer thread scheduling and reproduce elusive concurrency bugs. Ranked #118New way to see a problemPreprint4/10 suggested | LLMs for code via arxiv · 12 Oct, 16:29 | On-topic Uses an LLM agent reasoning over program semantics to detect and reproduce software concurrency bugs. | |
| Feature Competition and Its Representational Origins in Vision Language Models: Evidence from the Stroop Paradigm Benchmarks vision-language models on the Stroop paradigm to evaluate feature competition and text-versus-color bias. Ranked #118Surprising resultPreprint6/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Investigates feature competition and representational bias between text and color in vision-language models. | |
| An Investigation of Model Coherence: Narrow Finetunes Contradict Themselves Under Resampling Evaluates model coherence using 175 specific questions to measure internal contradictions and identity conflation under answer resampling. Ranked #118Surprising resultPreprint7/10 suggested | LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28 | On-topic Introduces a 175-question test and metric to evaluate model coherence and contradiction under resampling. | |
| Forgetting is Not Always Bad: A Neuro-Inspired Memory Repair Mechanism for Poisoned LLM Agents Introduces DREAM, an agent-intrinsic defense that triggers functional forgetting of poisoned memory entries through implicit reactivation and topological diagnosis. Ranked #118New way to see a problemPreprint7/10 suggested | AI agents via arxiv · 12 Oct, 16:28 | On-topic Designs memory repair mechanisms specifically for autonomous LLM agents. | |
| Reading the Neighborhood: Text-Attentive Graph Reranking for Automated Subject Indexing TAGHN enhances automated subject indexing by reranking candidate descriptors using a text-attentive graph over thesaurus neighborhoods. Ranked #119Preprint0/10 suggested | Retrieval-augmented generation (RAG) via arxiv · 15 Oct, 15:06 | On-topic Develops a graph reranker over retrieved candidate descriptors and document context. | |
| BRQGen: Towards Adaptive Completeness Assessment and Follow-Up Question Generation for Bug Reports The paper proposes a framework that identifies missing information in bug reports and generates targeted follow-up questions only when needed. Ranked #119New way to see a problemPreprint4/10 suggested | LLMs for code via arxiv · 15 Oct, 06:19 | On-topic Leverages LLMs to evaluate bug reports and iteratively generate follow-up questions for reproduction. | |
| RIFT: Token Influence Follows the Path of Least Resistance This paper proposes RIFT, a method using electrical resistance analogies to trace token influence through transformer attention graphs. Ranked #119New way to see a problemPreprint6/10 suggested | Small and efficient models via arxiv · 15 Oct, 06:18 | On-topic Proposes a fast, training-free method to compute token influence on a single CPU thread. | |
| Should SLMs be in the Driver's Seat? A Study of Local Models as Interpreters in Cloud-Written Agent Programs The paper demonstrates a hybrid architecture where cloud-generated agent code delegates localized private-data queries to edge models. Ranked #119New way to see a problemPreprint7/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:17 | On-topic Protects private data during deployment by delegating runtime interpretation to local edge models. | |
| CASTER: Formalizing Zero-Shot Manipulation from Synthetic Videos as Constrained Optimization Formulates zero-shot robot manipulation from synthetic video generation as a constrained trajectory optimization problem. Ranked #119New way to see a problemPreprint6/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Uses multimodal video generation models to guide real-world robot manipulation policies. | |
| Can LLMs Fix It Without Code? Toward Automated Verification of No-Code Bug Fixes Presents an execution-based evaluation pipeline using browser agents to verify language model solutions for no-code software bug reports. Ranked #119New testPreprint7/10 suggested | LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28 | On-topic Introduces an automated execution-based benchmark evaluating LLM capabilities on no-code bug fixes. | |
| Inject a Turn, Not a Token: Learning Where to Redirect LLM Reasoning for Self-Correction Introduces UPTURN, a framework that trains a small classifier to inject redirection prompts as new user turns during reasoning failures. Ranked #119Surprising resultPreprint7/10 suggested | AI agents via arxiv · 12 Oct, 16:28 | On-topic Designs a mechanism to intervene and redirect reasoning rollouts during multi-step problem solving. | |
| Decoding What iOS Location Services Send to Apple: An LLM Is an Unreliable Analyst but a Useful Toolmaker The study evaluates using language models to directly reconstruct mobile network schemas versus generating validation code tools. Ranked #120Surprising resultPreprint7/10 suggested | AI safety and reliability in deployment via arxiv · 17 Oct, 06:17 | On-topic Demonstrates LLM unreliability and subtle errors during reverse-engineering analysis. | |
| RCER: Reconstructing Cross-Evidence Relations for Iterative Retrieval-Augmented Generation RCER reconstructs cross-evidence relations and entity dependencies directly across iterative retrieval rounds to support complex multi-hop question answering. Ranked #120Preprint0/10 suggested | Retrieval-augmented generation (RAG) via arxiv · 15 Oct, 15:06 | On-topic Proposes an iterative retrieval-augmented generation method reconstructing cross-evidence relations. | |
| Hashrope: A Persistent Data Structure for Mechanisms of LLM Context Management This paper presents Hashrope, a unified persistent data structure supporting incremental edits, branching, and cache reuse for language model serving. Ranked #120Surprising resultPreprint6/10 suggested | Small and efficient models via arxiv · 15 Oct, 06:18 | On-topic Introduces an efficient data structure for prefix reuse, KV-cache management, and faster serving. | |
| Aligning Faithfulness for Theory-Scale Autoformalization The authors propose evaluating autoformalization faithfulness by aligning phrase-level correspondences between informal mathematics and formal code under contextual interpretations. Ranked #120New testPreprint4/10 suggested | LLMs for code via arxiv · 15 Oct, 06:17 | On-topic Evaluates autoformalization translating natural language mathematics into formal code. | |
| OneSearch-VL: Unified Multimodal Deep Research Agent for Image and Video Introduces a unified multimodal agent using evidence graphs to supervise external retrieval and reasoning across images and videos. Ranked #120New way to see a problemPreprint6/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Proposes a multimodal research agent linking visual grounding and retrieval across images and video. | |
| When VLMs Destroy What They Know: Spatial Representations Fall Below Lexical Priors The paper evaluates vision-language models' spatial representations using causally controlled perturbations, showing spatial relations degrade compared to lexical features. Ranked #120Surprising resultPreprint7/10 suggested | LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28 | On-topic Introduces a new metric and benchmark to evaluate spatial representation fidelity in VLMs. | |
| VEPA: Verified Evolving Formal Policy for Lifelong LLM Agent Guardrails Develops a framework for evolving deterministic formal policies to serve as runtime guardrails for large language model agents. Ranked #120New way to see a problemPreprint7/10 suggested | AI agents via arxiv · 12 Oct, 16:28 | On-topic Proposes formal evolving guardrails to regulate LLM agent actions over sensitive system resources. | |
| ReviewPilot: A Multi-Agent System for Screening and Data Extraction in Systematic Reviews Presents a multi-agent system for paper screening and data extraction in systematic literature reviews following standard guidelines. Ranked #121Preprint0/10 suggested | Retrieval-augmented generation (RAG) via arxiv · 17 Oct, 06:17 | On-topic Focuses on extracting, grounding, and citing evidence from academic literature databases. | |
| PAIR: Memory-Efficient and Fast Inference with Reading--Reasoning Fusion PAIR splits context processing and answer reasoning between two different-sized models, fusing predictions at the logit level for efficient inference. Ranked #121New way to see a problemPreprint6/10 suggested | Small and efficient models via arxiv · 15 Oct, 06:18 | On-topic Decouples reading and reasoning across different model sizes to cut KV memory and boost throughput. | |
| Nudge the Judge: Uncovering Program Decision Hijacking in Real-World Jev Applications The paper presents an automated framework to evaluate how adversarial contexts can redirect program execution through valid structured judgments in language-model applications. Ranked #121New testPreprint7/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:17 | On-topic Uncovers program decision hijacking attacks where malicious inputs redirect downstream software execution. | |
| Reasoning-Informed Visual Editing Introduces a multi-dimensional benchmark evaluating reasoning-informed visual editing across temporal, causal, spatial, and logical operations. Ranked #121New testPreprint6/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Introduces a benchmark evaluating instruction reasoning and appearance consistency in multimodal models. | |
| Measuring Artifact Reliance in Machine-Generated Text Detectors The paper demonstrates that machine-generated text detectors rely heavily on document formatting and PDF extraction artifacts rather than text quality. Ranked #121Surprising resultPreprint7/10 suggested | LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28 | On-topic Identifies formatting artifacts and introduces format counterfactuals for evaluating text detectors. | |
| QTestBench: A Test Framework for Evaluating LLM-Generated Quantum Code The paper introduces a benchmark framework evaluating LLM-generated quantum software unit tests against independent ground-truth checkers. Ranked #121New testPreprint3/10 suggested | LLMs for code via arxiv · 12 Oct, 16:28 | On-topic Examines LLMs generating unit test suites and evaluating code solutions for quantum software. | |
| Hindsight Structured Advice Distillation: Steering Executor Agents with Advisors Trained on Their Failures and Task Characteristics Trains an advisor model on past failures to add dynamic, structured guidance to frozen language model agent prompts. Ranked #121New way to see a problemPreprint7/10 suggested | AI agents via arxiv · 12 Oct, 16:28 | On-topic Proposes training an advisor model to guide executor agents on complex multi-step tasks. | |
| QEDIA: Quantum Program Verification in Lean Implements a formal framework in Lean 4 to verify hybrid quantum and classical programs. Ranked #122New way to see a problemPreprint3/10 suggested | LLMs for code via arxiv · 15 Oct, 06:18 | On-topic Uses an LLM-based prover agent to construct verified quantum programs. | |
| Does System One Transfer? Calibrated Typed-Decision Models, Label Budgets, and Translation for Indonesian Regional Languages The paper evaluates the cross-lingual transfer, calibration, and translation sensitivity of fast typed-decision models on regional Indonesian languages. Ranked #122Surprising resultPreprint6/10 suggested | Small and efficient models via arxiv · 15 Oct, 06:18 | On-topic Explores cheap, efficient calibrated decision models and confidence-gated routing to cut LLM costs. | |
| DPOison-VL: Clean Preference Poisoning in Vision-Language Direct Preference Optimization The paper introduces an imperceptible image perturbation attack that manipulates alignment in vision-language models during preference optimization without altering text annotations. Ranked #122Surprising resultPreprint7/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:17 | On-topic Presents clean preference poisoning attacks manipulating alignment in multimodal models. | |
| Towards Unified Evaluation of Prompt Enhancers for Video Generation Introduces PEBench, a benchmark and evidence-grounded framework to directly evaluate prompt enhancers for video generation without rendering full videos. Ranked #122New testPreprint6/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Presents an evaluation benchmark for prompt enhancers spanning text-to-video and image-to-video generation. | |
| Same Outcome, Different Evidence: Intent Recovery in LLM Safety Evaluation Pairs attack success rate with operative understanding rate to assess whether models recognize harmful intent during safety evaluations. Ranked #122Surprising resultPreprint7/10 suggested | LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28 | On-topic Critiques attack success rate and proposes operative understanding rate for LLM safety evaluation. | |
| SmartSalesBench: A Resistance-Aware Multi-Turn Benchmark for Evaluating LLM Sales Agents Introduces a multi-turn conversational benchmark to evaluate sales agents against resistant, psychologically grounded buyer personas. Ranked #122New testPreprint7/10 suggested | AI agents via arxiv · 12 Oct, 16:28 | On-topic Evaluates multi-turn autonomous LLM sales agents negotiating against resistant personas. | |
| MLExpert: Towards a Generalist Agentic Model for Autonomous Machine Learning Engineering The authors propose a post-training method enabling autonomous language model agents to organize experiments and carry out machine learning engineering tasks. Ranked #123Preprint0/10 suggested | LLMs for code via arxiv · 17 Oct, 06:17 | On-topic Focuses on training agentic models for code generation and software execution in ML engineering. | |
| AgentStripes: An Evidence-Graded Maturity Model for AI Agents The paper introduces a maturity model that grades artificial intelligence agents across multiple capability levels, evidence stages, and safety gates. Ranked #123New way to see a problemPreprint7/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:17 | On-topic Focuses directly on safety gates, controls, and reliability standards for deploying AI agents. | |
| Perception Test 2026: Challenge Summary and Extension to City-scale Audio-Visual Reasoning The paper introduces benchmarks and summarizes winning solutions for city-scale audio-visual reasoning and spatial intelligence at the Perception Test 2026. Ranked #123New testPreprint6/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Focuses on audio-visual video reasoning and multimodal models. | |
| Do Prompt Optimizers Break Requirements They Are Not Scored On? A Preregistered Study Evaluates whether automated prompt optimizers inadvertently cause regressions in unscored prompt constraints and requirements. Ranked #123Surprising resultPreprint7/10 suggested | LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28 | On-topic Evaluates prompt optimization benchmarks and assesses metrics on unmeasured constraints. | |
| Collaborative Theory of Mind: Learning to Self-Revise by Anticipating User Responses Trains language models to anticipate user replies and self-revise drafts before responding in multi-turn interactions. Ranked #123Surprising resultPreprint7/10 suggested | AI agents via arxiv · 12 Oct, 16:28 | On-topic Trains conversational agents to anticipate responses and self-revise during multi-turn interactions. | |
| ReplaySSM: Cache SSM Inputs, Not State Replaces frequent state memory updates in state-space models by caching inputs and periodically checkpointing state during decoding. Ranked #123New way to see a problemPreprint6/10 suggested | Small and efficient models via arxiv · 12 Oct, 16:28 | On-topic Accelerates decoding and cuts memory traffic for state-space models during inference serving. | |
| How Safe Is ChatGPT Health Advice? Evidence from Real User Conversations in Four Global South Countries Evaluates the safety of ChatGPT health advice using real user conversation logs across four Global South countries. Ranked #124Surprising resultPreprint7/10 suggested | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17 | On-topic Presents a demographically grounded safety evaluation rubric and benchmark using real-world user conversations. | |
| Configuring RAG Systems: What Breaks, Where Originates, and How to Fix The paper presents an empirical study of 654 real-world configuration bugs across ten retrieval-augmented generation systems, categorizing causes and fixes. Ranked #124New way to see a problemPreprint7/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:17 | On-topic Analyzes real-world deployment failures and configuration reliability in RAG systems. | |
| Look Where You Can: Active View Selection for CAD Reconstruction under Occlusion The paper introduces SightCAD, an active view selection framework optimizing camera viewpoints under occlusion constraints for parametric CAD reconstruction. Ranked #124New way to see a problemPreprint6/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Uses a vision-language model paired with visual view selection for CadQuery code generation. | |
| Meta-Reflective Evolution Effectively Improves Production AI Systems Introduces an evolutionary optimization method using multi-turn reflective mutators to improve production data agents and pipelines. Ranked #124Preprint0/10 suggested | LLMs for code via arxiv · 12 Oct, 16:28 | On-topic Uses an agentic code mutator to optimize production programs, data pipelines, and workflows. | |
| Locate, Judge, Read: Role-Specialized Tool Agents over a Knowledge-Point Index for Open-Domain Multimodal Document QA Designs a modular multi-agent pipeline separating search, verification, and reading for multimodal document question answering. Ranked #124Surprising resultPreprint7/10 suggested | AI agents via arxiv · 12 Oct, 16:28 | On-topic Designs specialized multi-agent roles and tool coordination to accomplish complex document retrieval tasks. | |
| DualQuant: Dual-scale Calibration-free Weight-and-Activation Quantization for LLMs Proposes a calibration-free post-training quantization method that jointly optimizes weight and activation scaling for 4-bit model inference. Ranked #124New way to see a problemPreprint6/10 suggested | Small and efficient models via arxiv · 12 Oct, 16:28 | On-topic Presents a calibration-free 4-bit weight-and-activation quantization method for efficient LLM serving. | |
| When Does an ASR Error Matter? Task-Role Diagnostics for Tool-Using Voice Agents The paper introduces a diagnostic protocol to evaluate how speech recognition errors affect structured tool calling in voice agents. Ranked #125New testPreprint7/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:17 | On-topic Evaluates the operational robustness and failure modes of deployed tool-using voice agents. | |
| ObliVul: Alert-Conditioned Safety Obligation Modeling and Bidirectional Counterfactual Validation for Code Vulnerability Detection Presents ObliVul, a framework combining safety obligation modeling and counterfactual validation to filter false static analysis vulnerability alerts. Ranked #125Preprint0/10 suggested | LLMs for code via arxiv · 12 Oct, 16:29 | On-topic Uses LLMs alongside program graphs to detect and validate code vulnerabilities. | |
| Seek-and-View Reasoning for Multi-View Spatial Understanding The paper proposes Vantage, a training-free framework pairing VLMs with 3D models to plan and synthesize relevant viewpoints for spatial reasoning. Ranked #125New way to see a problemPreprint6/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Enhances vision-language models for multi-view 3D spatial reasoning. | |
| Reflective Prompt Optimization of a Deep-Research Agent: A Topic-Disjoint, Multi-Seed Protocol for Measuring Transferable Gains and Prompt-Level Overfitting Studies reflective prompt optimization in research agents, demonstrating that gains on held-out tasks can stem from prompt-level memorization. Ranked #125Surprising resultPreprint7/10 suggested | LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28 | On-topic Proposes a rigorous topic-disjoint evaluation protocol to diagnose prompt-level overfitting. | |
| When Faster Decisions Do Not Make Better Agents: An Empirical Study of Jev in Agent Workflows Studies how accelerated decision-making models affect overall task completion, latency, and costs across embodied and mobile agents. Ranked #125Surprising resultPreprint7/10 suggested | AI agents via arxiv · 12 Oct, 16:28 | On-topic Directly evaluates decision-level acceleration techniques within multi-step autonomous agent workflows. | |
| A Privileged Teacher That Only Scores Cannot Search: Critique-Conditioned On-Policy Self-Distillation for Reasoning Examines error localization and credit assignment failures in critique-conditioned on-policy self-distillation for mathematical reasoning. Ranked #125Surprising resultPreprint6/10 suggested | Small and efficient models via arxiv · 12 Oct, 16:28 | On-topic Examines on-policy self-distillation techniques to train and improve reasoning models. | |
| Multi-Target Few-Shot Prompting with Diffusion Language Models Uses diffusion language models to perform multi-target few-shot infilling across shared code context guided by demonstration examples. Ranked #126Preprint0/10 suggested | LLMs for code via arxiv · 17 Oct, 06:17 | On-topic Applies multi-target diffusion prompting to software refactoring and code editing. |
841–900 of 3,333← PreviousNext →