Every paper checked — nothing hidden
All papers
Every paper checked for brief #4 of 12–19 Oct 2026 — nothing is hidden. Brief papers first.
All17,313In brief24More (on-topic)3,333Not relevant13,754Not assessed87Older352Could not confirm93Blogs with no paper266All blogs284
| Paper | Topic | Relevance | Feedback |
|---|---|---|---|
| Learning after Execution with Action-Specific Credit The paper proposes a reinforcement learning method that assigns distinct credit to initial code generation and subsequent post-execution correction steps. Ranked #485New way to see a problemPreprint5/10 suggested | AI agents via arxiv · 15 Oct, 06:17 | On-topic Explores learning after execution to improve sequential policy actions and corrective steps. | |
| A Better Metric Will Not Save You: What You Compare Beats How You Compare It Analyzes representational similarity metrics and demonstrates that input category selection has a larger impact on comparisons than the choice of metric. Ranked #485Surprising resultPreprint4/10 suggested | LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28 | On-topic Analyzes how comparative evaluation metrics and benchmark conditions impact model rankings. | |
| From Pixels to Structure: Lightweight Vision-Language Models for Document OCR and Structured JSON Extraction Evaluates eight lightweight open-source vision-language models on optical character recognition and schema-compliant JSON extraction across heritage document collections. Ranked #485Peer-reviewed0/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Directly evaluates vision-language models for document image understanding and structured extraction. | |
| L3M: A Verifiable Harness and Runtime for Self-Extending and Self-Improving Agents The paper presents a Lean 4 runtime allowing self-extending agents to run verified computations across tools without constant human oversight. Ranked #486New way to see a problemPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 15:06 | On-topic Develops a verifiable harness in Lean 4 to ensure reliable execution invariants in agent deployments. | |
| D-World: An Explicit World-Language-Action Model with Recursive Self-Improvement under Dual Control Designs an explicit digital-twin world model where robotic agents inspect, simulate, and calibrate physical parameters before acting. Ranked #486New way to see a problemPreprint5/10 suggested | AI agents via arxiv · 15 Oct, 06:17 | On-topic Proposes an LLM-based agent framework for robotic planning, digital twins, and iterative self-improvement. | |
| Why Heavy-Tailed Weight Spectra Predict Model Quality Provides a probabilistic marginal likelihood derivation explaining why heavy-tailed weight spectral metrics correlate with deep network generalization quality. Ranked #486New way to see a problemPreprint4/10 suggested | LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29 | On-topic Analyzes why heavy-tailed weight spectra theoretically predict model quality in DNNs and LLMs. | |
| PAGE: Parts-Aware Guidance for Co-Speech Gesture Portrait Video Generation Presents parts-specific reweighting and classifier guidance to improve facial details and audio alignment in speech-driven gesture videos. Ranked #486Peer-reviewed0/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Addresses cross-modal alignment between audio speech signals and visual video generation. | |
| Who's Behind the Text? Robust Model Authorship Attribution The paper proposes a likelihood-based attribution method to identify the source model of machine-generated text under distribution shifts. Ranked #487New testPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 15:06 | On-topic Addresses model authorship attribution to prevent misuse, cyberattacks, and unauthorized distillation in deployment. | |
| Scale, Drivers, and Limits of Representational Alignment Across Large Language Models The paper presents two geometric methods to evaluate where and why representational structures align across different large language models. Ranked #487New testPreprint4/10 suggested | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17 | On-topic Introduces new methods (NW-RSA and CF-RSA) to evaluate and compare representations across language models. | |
| Trace-to-Architecture: Capability Bottleneck Diagnosis for Harness Revision The paper introduces Trace-to-Architecture, a framework that diagnoses LLM agent execution failures to automatically revise agent harness components. Ranked #487New way to see a problemPreprint5/10 suggested | AI agents via arxiv · 15 Oct, 06:17 | On-topic Directly focuses on diagnosing execution bottlenecks and revising LLM agent harnesses. | |
| QK-Edit: Revisiting Attention-based Injection in MM-DiT for Image and Video Editing Introduces QK-Edit, a training-free query-key attention manipulation method for text-guided image and video editing in multimodal diffusion transformers. Ranked #487Peer-reviewed0/10 suggested | Multimodal models via arxiv · 15 Oct, 06:17 | On-topic Manipulates attention dynamics in multimodal diffusion transformers for text-guided image and video editing. | |
| REWIND: Uncertainty-Guided Trajectory Repair for Test-Time Reasoning The paper introduces a training-free controller that uses token entropy to target and regenerate faulty reasoning steps at test time. Ranked #488Surprising resultPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 15:06 | On-topic Addresses hallucination and error correction in deployment by dynamically repairing flawed reasoning paths. | |
| KARAMEL: Knowledge-Aware Ranking for Multimodal Entity Linking in the Arts via Heterogeneous Graphs The paper introduces a multimodal entity linking method that combines vision-language models with heterogeneous graph networks for cultural heritage data. Ranked #488Peer-reviewed0/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Fuses vision-language encoders with knowledge graphs for multimodal entity linking. | |
| SceneForesee: Learning Scene Evolution for Spatiotemporal Reasoning The paper introduces a benchmark and framework to evaluate how multimodal models track entities and reason about future scene evolution in videos. Ranked #488New testPreprint4/10 suggested | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17 | On-topic Introduces SceneForesee, a benchmark for evaluating spatiotemporal reasoning and entity continuity in vision models. | |
| CryptoFlowBench: Benchmarking Multi-Hop Reasoning on Blockchains for Cryptocurrency Fund Tracing Introduces a benchmark and tool-augmented agent for evaluating multi-hop reasoning in cryptocurrency fund tracing. Ranked #488New testPreprint5/10 suggested | AI agents via arxiv · 15 Oct, 06:17 | On-topic Introduces a graph-tool-augmented agent that iteratively reasons and traces multi-hop cryptocurrency flows. | |
| DANGERLENS: Probing Dangerous Representations Hidden by Safety Alignment The paper introduces an internal probing method to detect whether safety-aligned language models still retain accessible representations of hazardous concepts. Ranked #489Surprising resultPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 15:06 | On-topic Audits internal hazardous representations to measure whether safety alignment reliably neutralizes risk. | |
| Rethinking Supervision in Document Unwarping: A Self-consistent Flow-free Approach The paper introduces a flow-free deep network for smartphone document image unwarping using geometric consistency and undistorted document supervision. Ranked #489Peer-reviewed0/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Improves document understanding pipelines through rectification of smartphone document images. | |
| Do you "see" what I "see"? A Multi-panel Visual Question Answer Dataset for Large Language Model Chart Analysis This paper presents a synthetic multi-panel chart visual question answering dataset to evaluate how multimodal models interpret complex scientific figures. Ranked #489New testPreprint4/10 suggested | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17 | On-topic Introduces a new visual question answering benchmark for testing chart analysis capabilities. | |
| PopGen: Generating Semantically Diverse Populations for Zero-Shot Coordination The paper uses language models to generate semantically diverse agent policies in code to train reinforcement learning cooperators for zero-shot collaboration. Ranked #489New way to see a problemPreprint5/10 suggested | AI agents via arxiv · 15 Oct, 06:17 | On-topic Uses LLMs to generate diverse behavioral agent populations for multi-agent coordination. | |
| Tail Competition Explains Scaling in Best-of-N Sampling for Verifiable Problems The paper analyzes Best-of-N sampling as an extreme-value competition between tails of correct and incorrect responses to explain reward hacking. Ranked #490New way to see a problemPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 15:06 | On-topic Explains how reward hacking causes degradation and introduces a criterion to prevent harmful sampling outcomes. | |
| Bidirectional Adaptive Transformers for Multimodal Anomaly Detection Proposes a bidirectional adaptive transformer framework to align RGB and 3D point cloud features for multimodal anomaly detection. Ranked #490Peer-reviewed0/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Addresses multimodal alignment and fusion between 2D RGB images and 3D point clouds. | |
| Imitation Holds a Key to Evolution: Localized Repair and Verified Skill Acquisition for Adaptive Agents This paper presents an imitation framework that enables agents to identify violated contracts, extract skills, and repair execution failures. Ranked #490New way to see a problemPreprint5/10 suggested | AI agents via arxiv · 15 Oct, 06:17 | On-topic Presents a controller for adaptive agents to localize failures and acquire skills. | |
| Not All Drift Is a Mechanism: Tracing Prompt Adaptation in LLMs The authors trace internal representations across prompt optimization techniques to show how discrete prompts, prompt tuning, and prefixes alter model processing. Ranked #490New way to see a problemPreprint4/10 suggested | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17 | On-topic Examines whether internal representation measurements reliably evaluate prompt adaptation. | |
| GeoState: Learning View-Decoupled Geometry States for 3D World Modeling The authors introduce GeoState, a view-decoupled 3D world modeling approach separating scene dynamics from camera motion. Ranked #491Preprint0/10 suggested | Multimodal models via arxiv · 17 Oct, 06:17 | On-topic Develops a 3D geometry representation model operating on multi-view visual inputs. | |
| Can We Believe AI Peer Review? The paper evaluates the reliability, score variance, and consistency of automated peer reviews generated by large language models. Ranked #491Surprising resultPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 15:06 | On-topic Assesses reliability and inconsistency risks of deploying LLM scoring in peer review pipelines. | |
| Online LLM Personalization from Pairwise Preferences Across Users The paper formulates online language model persona personalization as a dueling bandit problem using sequential pairwise user preferences. Ranked #491New testPreprint4/10 suggested | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17 | On-topic Releases an online preference benchmark and dataset to evaluate personalization algorithms. | |
| Heterogeneous Agentic Coevolution This paper proposes a coevolution framework where language model agents and domain foundation models iteratively train and improve each other. Ranked #491New way to see a problemPreprint5/10 suggested | AI agents via arxiv · 15 Oct, 06:17 | On-topic Proposes coevolutionary reinforcement learning framework to improve multi-step agent planning and specialist models. | |
| TimeCap: A Scalable Open-Source Pipeline for Captioning Wearable Health Time-Series Data The paper introduces TimeCap, an open deterministic pipeline for generating signal-grounded captions from wearable health time-series data. Ranked #492Preprint0/10 suggested | Multimodal models via arxiv · 17 Oct, 06:17 | On-topic Focuses on cross-modal pairing of sensor time-series data with natural language. | |
| Whose Data, Whose Benchmark? A User-Centric Framework for Auditing Evaluation Resources The paper introduces a framework to audit cultural, geographic, and user-representation biases embedded within popular machine learning benchmarks. Ranked #492New way to see a problemPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 15:06 | On-topic Examines reliability, user assumptions, and measurement validity in model evaluation resources. | |
| GENDIS: Gender-Related Discourse and Stance in Multilingual YouTube Comments from India, Kenya, and Nigeria The paper presents GENDIS, a multilingual dataset of YouTube comments from India, Kenya, and Nigeria annotated for gender-related discourse and stance. Ranked #492New testPreprint4/10 suggested | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17 | On-topic Introduces a new multilingual dataset and benchmark for evaluating gender-related discourse and stance detection. | |
| CB-Agent: Concept-Based In-Context Learning for Sequential Action Selection The paper presents an in-context learning method that extracts compact concept vectors from demonstrations to guide sequential action choices for agents. Ranked #492New way to see a problemPreprint5/10 suggested | AI agents via arxiv · 15 Oct, 06:17 | On-topic Guides sequential action selection in interactive ReAct agents using concept-based in-context learning. | |
| Bias Mitigation in Audio-Enabled LLMs through Adversarial Training and Distillation Methods Applies adversarial training and distillation to remove unwanted gender bias in audio-enabled language models. Ranked #493Preprint0/10 suggested | Multimodal models via arxiv · 17 Oct, 06:17 | On-topic Directly focuses on audio-enabled multimodal large language models. | |
| Identity, Incentives, and Cooperative Behavior for AI: Reverse-Engineering Multiagent Cooperation The paper develops an analytical framework to interpret and reverse-engineer incentives and cooperative behaviors in multi-agent artificial intelligence systems. Ranked #493New way to see a problemPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 15:06 | On-topic Investigates misaligned collective behaviors and mechanisms driving multi-agent coordination safety risks. | |
| What and Where in Identifying and Quantifying Conventions in Humans and AI Agents The paper analyzes implicit communication conventions between humans and AI agents in the cooperative card game Hanabi. Ranked #493New way to see a problemPreprint5/10 suggested | AI agents via arxiv · 15 Oct, 15:06 | On-topic Studies cooperative conventions and communication in multi-agent game interactions. | |
| AD4EDU: Benchmarking University Lecture Audio Description across Accessibility Standards, Metrics, and Learner Preferences This paper introduces a benchmark and rule-based metric for generating and evaluating audio descriptions in university slide lectures. Ranked #493New testPreprint4/10 suggested | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17 | On-topic Introduces AD4Edu-Bench and an evaluation metric for assessing audio descriptions. | |
| Enhancing Multimodal RLVR with Pairwise Token Contrast and Adaptive Trajectory Confidence Improves reinforcement learning credit assignment in multimodal reasoning using pairwise token contrast and adaptive trajectory confidence. Ranked #494Preprint0/10 suggested | Multimodal models via arxiv · 17 Oct, 06:17 | On-topic Improves multimodal reasoning by isolating image dependence during reinforcement learning. | |
| MolLean: A Process-Verifiable Language for Chemical Reasoning This paper introduces MolLean, a verifiable language and tool interface that grounds chemical reasoning in formal state transitions and explicit checks. Ranked #494New way to see a problemPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 15:06 | On-topic Reduces hallucinations in chemical reasoning via process verification and kernel checks. | |
| CASI: Credit Assignment via Set Intervention for Agentic Retrieval The paper presents an intervention method using Shapley values to assign credit to individual actions within multi-action agentic retrieval sets. Ranked #494New way to see a problemPreprint5/10 suggested | AI agents via arxiv · 15 Oct, 15:06 | On-topic Addresses credit assignment for agents executing multiple retrieval actions jointly. | |
| Attention-Graph Curvature for the Verification of Mathematical Reasoning This paper evaluates whether attention-graph curvature features can help verifiers detect errors in mathematical reasoning steps. Ranked #494New way to see a problemPreprint4/10 suggested | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17 | On-topic Evaluates per-step verifiers for detecting errors in complex mathematical reasoning sequences. | |
| Task-Centric Multimodal Learning for Cognitive Impairment Detection Evaluates a multimodal screening method using voice, video, and text from voice-assistant interactions to detect cognitive impairment. Ranked #495Preprint0/10 suggested | Multimodal models via arxiv · 17 Oct, 06:17 | On-topic Evaluates multimodal fusion across video, audio, and text for cognitive assessment. | |
| Same Claim, Different Interpretations: How LLM Agents Resolve Ambiguity in Scientific Claims This paper proposes Claim Cards to make agent interpretations explicit when verifying ambiguous natural-language scientific claims. Ranked #495New way to see a problemPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 15:06 | On-topic Addresses reliability and hidden interpretation divergence in deployed LLM scientific verifiers. | |
| Price of Anarchy of First-Price Auctions in Agentic Markets Analyzes the price of anarchy and economic efficiency of first-price auctions where AI agents bid to serve user prompts. Ranked #495New way to see a problemPreprint5/10 suggested | AI agents via arxiv · 15 Oct, 15:06 | On-topic Analyzes market mechanisms and auction dynamics for decentralized serving-time LLM agents. | |
| EHR-Bench: A Chinese Multi-Task Benchmark for Recruitment AI This paper introduces a Chinese multi-task evaluation benchmark covering six recruitment application areas for language models. Ranked #495New testPreprint4/10 suggested | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17 | On-topic Introduces EHR-Bench, a multi-task benchmark and evaluation framework for recruitment AI. | |
| ACEdit: Stable Multi-Turn Image Editing via Asymmetric Conditioning The authors propose ACEdit, a training and distillation scheme that stabilizes image editing models against accumulating errors across multiple editing turns. Ranked #496Preprint0/10 suggested | Multimodal models via arxiv · 17 Oct, 06:17 | On-topic Investigates multi-turn vision-language image editing and instruction-following models. | |
| Sparse Autoencoders for Mechanistic Interpretability in NLP: A Critical Survey of Recoverability, Interpretability, and Utility This paper surveys 138 studies on sparse autoencoders in NLP, examining dictionary design, feature recoverability, and practical utility. Ranked #496New way to see a problemPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 15:06 | On-topic Surveys sparse autoencoders used to interpret and monitor internal model representations. | |
| Beyond LLM Judge: Mechanical Verification for Agentic Autoformalization Combines language model review with mechanical numerical checks to verify mathematical formalizations and catch errors missed by automated judges. Ranked #496New testPreprint5/10 suggested | AI agents via arxiv · 15 Oct, 15:06 | On-topic Proposes a verification pipeline for agentic autoformalization. | |
| Local Citation Recommendation Without Graph Signals The paper introduces a dataset and citation recommendation system that retrieves relevant scientific papers without relying on citation graphs. Ranked #496New testPreprint4/10 suggested | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18 | On-topic Introduces a new benchmark dataset for local citation recommendation. | |
| Twice Upon a Time: Complementary Representations for Time-Series Reasoning The authors propose combining line plots and normalized delay embeddings to provide complementary visual representations for time-series reasoning in language models. Ranked #497Preprint0/10 suggested | Multimodal models via arxiv · 17 Oct, 06:17 | On-topic Applies vision-language foundation models to visual time-series reasoning via dual visual representations. | |
| What Is a Clarification? Estimating Model-Dependent Input Ambiguity The paper formalizes clarification criteria to estimate model-dependent input ambiguity based on accessible and valid answers. Ranked #497New way to see a problemPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 15:06 | On-topic Helps detect ungrounded answers and ambiguity in models to improve reliability. | |
| Interlock: From Local Correctness to Task Completion in Long-Horizon Agents The paper proposes a graph-based task synthesis framework to help long-horizon agents satisfy inter-dependent constraints and achieve complete task outcomes. Ranked #497New way to see a problemPreprint5/10 suggested | AI agents via arxiv · 15 Oct, 15:06 | On-topic Develops a framework to evaluate and ensure task completion for long-horizon agents. | |
| How Reliable are AMR Parser Outputs? Uncertainty Estimation for AMR Parsing Studies uncertainty estimation for abstract meaning representation parsers using top-k beam candidates and prediction rejection ratios. Ranked #497New testPreprint4/10 suggested | LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28 | On-topic Explores uncertainty estimation and metric reliability for evaluating semantic parsers. | |
| Gray-Box Hallucination Detection in LLMs with Token Identities The paper introduces a lightweight gray-box hallucination detector combining token probabilities with vector representations of token identities. Ranked #498Surprising resultPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 15:06 | On-topic Presents a gray-box detection framework to catch model hallucinations in production. | |
| Euler: Towards Human-Centered Multi-Agent Mathematical Discovery This paper introduces a multi-agent mathematical discovery system that organizes proofs, assumptions, and failed attempts into structured research graphs. Ranked #498New way to see a problemPreprint5/10 suggested | AI agents via arxiv · 15 Oct, 15:06 | On-topic Develops a multi-agent system coordinating structured planning, collaboration, and proof verification. | |
| GeoReader: Learning Latent Geometry for Spatial Reasoning in Vision-Language Models This paper introduces a framework that trains lightweight readers to convert latent 3D geometric representations into tokens readable by vision-language models. Ranked #498Preprint0/10 suggested | Multimodal models via arxiv · 15 Oct, 06:17 | On-topic Improves 3D spatial reasoning in vision-language models by integrating latent geometric representations. | |
| GeoFitBench-LLM: A Benchmark for Assessing Large Language Models' Capabilities in Geospatial Data Requirements Elicitation for Geo-Analytic Tasks Introduces a benchmark evaluating language models on eliciting geospatial data requirements and constraints from natural language requests. Ranked #498New testPeer-reviewed4/10 suggested | LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28 | On-topic Proposes GeoFitBench-LLM, a benchmark evaluating LLM capabilities in geospatial requirement elicitation. | |
| SoK: Grading Open-Text Legal Answers, from Law School Exams to Trustworthy Rubric Judges The paper systematizes grading requirements for open-text legal exams and compares eight evaluation paradigms using human and model judges. Ranked #499New way to see a problemPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 15:06 | On-topic Examines reliability, failure modes, and gaming risks of LLM judges in high-stakes testing. | |
| VideoHarness-SRSI: Structured Recursive Self-Improvement for Video Understanding Harness The paper examines failure modes in recursively self-improving execution harnesses for agentic video understanding models. Ranked #499New way to see a problemPreprint5/10 suggested | AI agents via arxiv · 15 Oct, 15:06 | On-topic Designs recursive self-improvement harnesses for agentic video understanding workflows. | |
| Just Respect the Semantics: Image-Token Compression via Pre-LLM Semantic Variable-Resolution Tokenization and In-LLM Relevance Pruning The authors present an image-token compression technique using variable-resolution merging for semantic regions and learned relevance pruning inside the model. Ranked #499Preprint0/10 suggested | Multimodal models via arxiv · 15 Oct, 06:17 | On-topic Introduces token pruning and variable-resolution tokenization methods for multimodal vision-language models. | |
| Looking Inside LLMs: Small-World Connectivity as a Signature of Reasoning Performance Analyzes attention-head activation graphs to show that small-world connectivity correlates with reasoning performance in large language models. Ranked #499New way to see a problemPreprint4/10 suggested | LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28 | On-topic Investigates structural metrics that correlate with and evaluate LLM reasoning capabilities. | |
| GraphRAG-LTM: A Layered Threat Modeling Framework for Attacks on Graph Retrieval-Augmented Generation The paper introduces a threat modeling framework examining adversarial query demotion and graph edit attacks on graph-based RAG systems. Ranked #500New way to see a problemPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 15:06 | On-topic Investigates security vulnerabilities and adversarial attacks on GraphRAG systems in deployment. |
2,761–2,820 of 17,313← PreviousNext →