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 |
|---|---|---|---|
| RoLaQ: Joint Rotation and Range Calibration for 2-Bit KV Caches in Multi-Head Latent Attention This paper presents a calibration method to compress multi-head latent attention key-value caches to two bits for efficient long-context language model inference. Ranked #458Preprint0/10 suggested | Small and efficient models via arxiv · 17 Oct, 06:18 | On-topic Proposes rotation and range calibration for 2-bit KV cache compression to improve inference efficiency. | |
| SERO: Learning from Executable Oracle Continuations for Long-Horizon Agentic Reinforcement Learning The paper proposes an agentic reinforcement learning method that uses executable oracle continuations to guide exploration during long-horizon tasks. Ranked #458New way to see a problemPreprint5/10 suggested | AI agents via arxiv · 15 Oct, 15:06 | On-topic Introduces an RL framework using executable oracle continuations for long-horizon agent tasks. | |
| MiMaAlign: Hierarchical Micro-Macro Alignment for Physically Plausible Video Generation The paper proposes a hierarchical alignment method that transfers continuous dynamics and high-level event knowledge into video diffusion models. Ranked #458Preprint0/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Aligns text-to-video diffusion models using vision-language model event semantics. | |
| StableJudge: A Lightweight Adversarial Defense Framework for LLM-as-a-Judge The paper presents StableJudge, an external lightweight sentry model that sanitizes prompt-injection attacks before LLM-as-a-judge scoring. Ranked #458New way to see a problemPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:17 | On-topic Mitigates adversarial prompt-injection attacks on deployed LLM-as-a-judge evaluation pipelines. | |
| DuplexAct: How Interactive Are Frontier Full-Duplex Spoken Language Models? This paper presents a benchmark with an action taxonomy to evaluate the interactive conversation capabilities of full-duplex spoken language models. Ranked #458New testPreprint5/10 suggested | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17 | On-topic Introduces DuplexAct, a structured benchmark evaluating interaction actions in full-duplex spoken language models. | |
| GHOST: Geometry-Constrained Hybrid Diffusion for Open-Vocabulary 3D Scene Graph Generation The paper proposes a hybrid diffusion method that enforces geometric constraints to generate open-vocabulary 3D scene graphs from point clouds. Ranked #459Preprint0/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Connects open-vocabulary language embeddings with 3D spatial geometry in scene graph generation. | |
| Separating Prompt Harmfulness from Interaction Safety via Partial Generation The paper proposes evaluating partial generation outputs to stop harmful model interactions early while avoiding unnecessary prompt rejections. Ranked #459Surprising resultPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:17 | On-topic Introduces partial-generation guardrails to prevent harmful model responses during deployment. | |
| MMTA: Benchmarking Multimodal Temporal Analysis with Time Series, Text, and Vision Releases MMTA, a multimodal benchmark testing the ability of language models to jointly analyze time series, text, and visual data. Ranked #459New testPreprint5/10 suggested | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17 | On-topic Introduces a comprehensive benchmark evaluating multimodal models on time series, text, and visual data. | |
| JANUS: Enhancing Long-Horizon Autonomous Research with Co-Evolving World Models Proposes JANUS, using a co-evolving world model to predict, verify, and guide exploration paths in autonomous scientific research. Ranked #459New way to see a problemPreprint5/10 suggested | AI agents via arxiv · 12 Oct, 16:28 | On-topic Develops autonomous research agents using planning, world models, and reflective execution loops. | |
| From Sound to Sense: An Anatomy of Semantic Supervision for Speech-Driven Visual Grounding in Remote Sensing Imagery The paper analyzes semantic supervision mechanisms in direct speech-to-box grounding models for remote sensing imagery. Ranked #460Preprint0/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Investigates audio-visual multimodal grounding aligning speech features with satellite imagery. | |
| Dream2Discern: Learning from Hard Latent Variants for Runtime Failure Prediction in Multi-Agent Systems The paper presents a runtime auditor that predicts execution failures in multi-agent systems using loss-guided perturbations to identify hard failure cases. Ranked #460New way to see a problemPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:17 | On-topic Develops a runtime failure prediction and monitoring system for reliable multi-agent deployment. | |
| Know-Thy-Not: Training-Free Targeted Negation in Dual-Encoder Vision-Language Models The paper introduces a training-free technique and benchmark to help vision-language models correctly process and reject negated concepts in text prompts. Ranked #460New testPreprint5/10 suggested | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17 | On-topic Introduces the COCO-Confusion benchmark to evaluate negation in vision-language models. | |
| OpenSocialJax: Evaluating Social Adaptation in Sequential Social Dilemmas with Open-Ended Environmental Dynamics Introduces a benchmark evaluating how language model agents adapt to latent environmental dynamics and social dilemmas over repeated interactions. Ranked #460New testPreprint5/10 suggested | AI agents via arxiv · 12 Oct, 16:29 | On-topic Evaluates multi-agent LLM teams on social dilemmas, rule inference, and cooperation. | |
| Learning Temporal Evidence Selection for Video Anomaly Understanding The paper proposes a learned temporal evidence selection approach to generate grounded text explanations for video anomaly understanding. Ranked #461Preprint0/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Uses vision-language and Q-Former models to select visual evidence for language generation. | |
| Reels-RSI: Program-Level Self-Improvement for Code-Rendered Explainer Videos The paper introduces an open-source agent framework that improves code-rendered explainer videos through program-level debugging, static rules, and unit testing. Ranked #461New way to see a problemPreprint5/10 suggested | AI agents via arxiv · 15 Oct, 06:17 | On-topic Presents an agent harness that iteratively writes, checks, and improves animation code programs. | |
| Computational Reflective Equilibrium for Agentic AI: A Rawlsian Alignment Scaffold for Governance The paper proposes an ethical governance scaffold based on reflective equilibrium to resolve conflicting instructions and policy rules in agentic workflows. Ranked #461New way to see a problemPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:17 | On-topic Provides a Rawlsian governance harness enforcing prohibitions, provenance, and due diligence in agent actions. | |
| HelpMeOut: A Neuropsychology-Grounded Benchmark for Error-Aware Monitoring of Household Task Execution Introduces HelpMeOut, an embodied AI benchmark for diagnosing human execution errors across household activity trajectories motivated by neuropsychology research. Ranked #461New testPreprint5/10 suggested | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17 | On-topic Introduces HelpMeOut, a benchmark for evaluating error-aware monitoring of task execution. | |
| UMAS3D: Unified Mask-Free 3D Editing with Region-Adaptive Source Preservation The paper proposes a training-free 3D editing framework that estimates edit regions without masks to preserve unedited source geometry and appearance. Ranked #462Preprint0/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Presents a multimodal 3D asset editing framework conditioned on text, boxes, and images. | |
| SkillScaffold: Scaffolding Agentic Reinforcement Learning with Policy-Adaptive Skill Credit Assignment and Selective Injection This paper introduces a reinforcement learning scaffold that selectively injects textual skills into agents using low-cost utility estimation. Ranked #462New way to see a problemPreprint5/10 suggested | AI agents via arxiv · 15 Oct, 06:17 | On-topic Proposes scaffolding techniques to improve multi-turn agentic reinforcement learning. | |
| PID-Gated Continual Prompt Adaptation Under Noisy Feedback This paper proposes control-inspired gates to decide when language model agents should adapt their prompts without absorbing noisy user feedback. Ranked #462New way to see a problemPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:17 | On-topic Improves prompt adaptation stability and robustness under noisy feedback in deployment. | |
| Beyond Successful Play: Diagnosing Opponent Modeling in Large Language Models The paper introduces a diagnostic framework to test whether language model agents genuinely infer opponent beliefs rather than following superficial behavioral patterns. Ranked #462New testPreprint5/10 suggested | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17 | On-topic Presents a benchmark framework diagnosing theory of mind in language models. | |
| Structured Stateful Relational Scripts with a High-Quality Dataset for Multi-Shot Audio-Visual Video Generation The paper introduces structured stateful relational scripts and a paired dataset to coordinate persistent identities and audio events in multi-shot video generation. Ranked #463Preprint0/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Introduces representation and dataset for coordinated multimodal audio-visual video generation. | |
| PlanPath: Exploration-Guided Orchestration for Patient-Level Pathology Reasoning The paper proposes an agent framework that plans multi-slide exploration and visual tool use for patient-level pathology visual question answering. Ranked #463New way to see a problemPreprint5/10 suggested | AI agents via arxiv · 15 Oct, 06:18 | On-topic Presents an agent framework that plans and uses tools for cross-slide reasoning. | |
| RetryBench: Separating Retry Permission from Recovery Outcomes in GPT Tool Use The paper presents a benchmark evaluating how language model agents handle unauthorized tool call retries after receiving timeouts. Ranked #463New testPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:17 | On-topic Studies agent reliability failures when retrying unauthorized or unsupported tool calls. | |
| PatTransBench: an ipc-stratified, multi-dimensional benchmark for patent machine translation evaluation Presents an evaluation benchmark and multi-dimensional protocol for patent machine translation across Chinese, English, Japanese, and Korean. Ranked #463New testPreprint5/10 suggested | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17 | On-topic Introduces a stratified multi-dimensional benchmark and evaluation protocol for patent machine translation. | |
| OmniRSI: Taming Latent Diffusion for Unified Multimodal Remote Sensing Image Generation The paper proposes a unified latent diffusion architecture with a custom variational autoencoder for multimodal remote sensing image generation. Ranked #464Preprint0/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Presents a unified multimodal generative model conditioning on text and heterogeneous visual inputs. | |
| Engram Injection: A Study of Temporal Validity in Parametric Language-Model Memory The paper studies whether language models can track temporal validity changes in parametric memory via adapter conditioning and clocks. Ranked #464Surprising resultPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:17 | On-topic Investigates factual accuracy and temporal validity in model parametric memory to prevent hallucinations. | |
| 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 #464New testPreprint5/10 suggested | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17 | On-topic Presents a benchmark spanning multi-hop reasoning tasks for cryptocurrency transaction tracing. | |
| Delegating Earth science model calibration to knowledge-guided LLM agents The authors introduce a calibration harness using structured domain knowledge to safely delegate Earth science model parameterization to language model agents. Ranked #464New way to see a problemPreprint5/10 suggested | AI agents via arxiv · 15 Oct, 06:17 | On-topic Delegates complex scientific model calibration workflows to knowledge-guided LLM agents. | |
| Reference-Efficient Multimodal Bitstream-Corrupted Video Recovery via Depth-Conditioned Adaptation This paper presents a depth-conditioned video recovery framework that selects a single reliable RGB reference to restore damaged video bitstreams efficiently. Ranked #465Preprint0/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Combines RGB appearance and depth geometry to reconstruct corrupted video sequences. | |
| A Persian Vision-Language-Action Dataset and Benchmark for Ambiguity- and Infeasibility-Aware Instruction-to-Plan Grounding The paper introduces a Persian vision-language-action dataset and benchmark to evaluate robotic instruction grounding under deliberate ambiguity and infeasibility. Ranked #465New testPreprint5/10 suggested | AI agents via arxiv · 15 Oct, 06:18 | On-topic Studies agentic planning and disambiguation for embodied instruction-to-plan grounding. | |
| Knowledge Interaction in Hypernetwork-based Continual Learning: Conflict and Composition The paper examines how hypernetwork adapters resolve knowledge conflicts and compose information during continual learning in language models. Ranked #465Surprising resultPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:17 | On-topic Studies knowledge conflicts and updating reliability in continually adapted LLMs. | |
| ScreenASR-Bench: A Tiered Benchmark for Speech Recognition with Visual Context Introduces a tiered benchmark to assess how explicit and implicit visual evidence helps speech recognition models resolve ambiguous words on screens. Ranked #465New testPreprint5/10 suggested | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17 | On-topic Introduces ScreenASR-Bench to evaluate models across multimodal context tiers. | |
| SlotAVS: Object-Centric Audio-Visual Segmentation via Bi-Modal Slot Attention This paper introduces a competitive bi-modal slot attention method to cleanly segment distinct sounding objects in multi-source audio-visual scenes. Ranked #466Preprint0/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Proposes an object-centric model binding audio and visual signals for segmentation. | |
| GesBench: When and Where Do Multimodal Language Models Perceive Social Meaning? Introduces a benchmark evaluating how and when multimodal language models perceive human social cues from partial audio and video snippets. Ranked #466New testPreprint5/10 suggested | LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17 | On-topic Introduces GesBench, a new benchmark testing cue-level social perception in multimodal models. | |
| Accuracy and Uncertainty in Machine Deception Detection (Research White Paper) The paper proposes a multi-layer detector ensemble to identify deceptive model outputs while calibrating score uncertainty and false-positive rates. Ranked #466New way to see a problemPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:17 | On-topic Proposes activation monitoring to detect deceptive behavior with calibrated uncertainty bounds in deployment. | |
| Can Language Models Play the Wikipedia Game? Tests language models on graph-based information retrieval and navigation using the Wikipedia Game across multiple languages and citation datasets. Ranked #466New testPreprint5/10 suggested | AI agents via arxiv · 12 Oct, 16:28 | On-topic Tests LLM navigation and multi-step link-following decision making on graph structures. | |
| SITAR: A Situational Distortion Metric for Automatic Speech Recognition Evaluation Develops a lightweight situational distortion metric to evaluate speech recognition errors based on practical conversational severity. Ranked #467New testPreprint5/10 suggested | LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17 | On-topic Proposes a new situational distortion metric and validation protocol for evaluating speech transcription. | |
| Sparse All-Layer Connector for Domain Generalised Semantic Segmentation This paper introduces a sparse connector allowing cross-layer feature sharing across different foundation models for domain-generalized semantic segmentation. Ranked #467Preprint0/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Fuses representations from vision-language and vision foundation models across layers. | |
| UDHRBench: Benchmarking Human Rights in Large Language Models Evaluates commercial language models against Universal Declaration of Human Rights principles, revealing misalignments driven by efficiency arguments. Ranked #467New testPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:17 | On-topic Evaluates LLM alignment and systematic safety failures regarding human rights compliance. | |
| LAPT-Scale: Scaling Embodied Policies with Latent Action Pre-Training on Human Videos Pre-trains embodied robot policies on human egocentric videos using learned continuous latent action representations. Ranked #467New way to see a problemPreprint5/10 suggested | AI agents via arxiv · 12 Oct, 16:28 | On-topic Trains embodied vision-language-action policies using latent action pre-training on human videos. | |
| Learning When to Withdraw: Curriculum Coverage and Rule-Specific Failures in Fine-Tuned Language Agents Studies how fine-tuning helps tool-using agents learn to withdraw and repair actions when environment assumptions change. Ranked #468Surprising resultPreprint5/10 suggested | AI agents via arxiv · 17 Oct, 06:17 | On-topic Studies how fine-tuned language agents revise, suspend, and repair tool actions dynamically. | |
| Twenty Days Later: A Pre-Registered Test of History-Based Forecasts of Brand Mentions in Gemini Answers The paper reports a pre-registered test measuring the stability of history-based forecasts for model brand mentions over twenty days. Ranked #468Surprising resultPreprint5/10 suggested | LLM evaluation and benchmarks via arxiv · 15 Oct, 15:06 | On-topic Evaluates temporal reliability and forecast durability of LLM-generated brand mentions across multiple languages. | |
| CatPortrait: Taming Video Generative Models for Portrait Animation via In-context Conditioning The paper presents a framework that uses structured 3D facial guidance as in-context input conditioning for controllable portrait video animation. Ranked #468Preprint0/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Animates portraits using video and audio conditioning modalities. | |
| Trust Hinders Bidirectional Human-AI Alignment Critiques trust calibration in human-AI decision making for open-ended problems and outlines dimensions for higher-order system alignment. Ranked #468New way to see a problemPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:17 | On-topic Examines human-AI alignment, uncertainty management, and trust calibration in high-stakes decisions. | |
| Can LLMs Access their World Knowledge for Event Prediction? The authors introduce DescribeEvent to show that language models can forecast temporal events when using natural-language descriptions and symbolic labels. Ranked #469New way to see a problemPreprint5/10 suggested | LLM evaluation and benchmarks via arxiv · 15 Oct, 15:06 | On-topic Introduces the DescribeEvent benchmark to evaluate how language models predict temporal events. | |
| Text-Guided Medical Image Segmentation via Residual Aggregation and Dense Alignment Representations The paper presents a parameter-efficient architecture connecting text embeddings with mask decoding for language-guided medical image segmentation. Ranked #469Preprint0/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Grounds natural language text prompts into medical image segmentation. | |
| To Defer or Not to Defer? Online Learning with LLMs and Congested Experts Proposes an online learning algorithm that decides whether language models should answer queries directly or defer to congested human experts. Ranked #469New way to see a problemPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:17 | On-topic Studies selective query deferral under uncertain LLM confidence to ensure reliable deployment. | |
| MAGMA-Bench: Evaluating Long-Horizon Embodied Agents under Dynamic Tasks and Execution Failures Introduces MAGMA-Bench to evaluate long-horizon embodied agent reasoning during dynamic task execution failures and multi-robot coordination. Ranked #469New testPreprint5/10 suggested | AI agents via arxiv · 12 Oct, 16:28 | On-topic Focuses directly on multi-agent planning, coordination, and handling execution failures in dynamic embodied tasks. | |
| 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 #470New testPreprint5/10 suggested | LLM evaluation and benchmarks via arxiv · 15 Oct, 15:06 | On-topic Introduces mechanical verification to replace LLM judges in formal verification. | |
| [Re] Boosting the Visual Interpretability of CLIP via Adversarial Fine-Tuning The paper presents a reproducibility study and an explicit saliency supervision extension for adversarial fine-tuning of CLIP vision models. Ranked #470Preprint0/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Focuses on visual interpretability and concept alignment in vision-language CLIP models. | |
| What Survives When Activation Steering Is Used as Memory This paper evaluates the capabilities and limitations of using activation steering vectors as an agent memory mechanism. Ranked #470Surprising resultPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:17 | On-topic Demonstrates reliability limits and preference reversal risks when using activation steering for memory. | |
| RoleWorld: Knowledge Transfer in World Modeling with Functional Role Abstractions Proposes a symbolic world modeling method using functional role abstractions to transfer interaction dynamics across environments for LLM agents. Ranked #470New way to see a problemPreprint5/10 suggested | AI agents via arxiv · 12 Oct, 16:28 | On-topic Designs world modeling abstractions to improve LLM agent decision-making and action anticipation. | |
| PerSo: Investigating Facets of LLMs' Social Intelligence in Real-World Persuasion Scenarios Introduces a benchmark based on real online discussions to evaluate both cognitive and behavioral persuasion skills in language models. Ranked #471New testPreprint5/10 suggested | LLM evaluation and benchmarks via arxiv · 15 Oct, 15:06 | On-topic Introduces a new benchmark evaluating social and cognitive intelligence. | |
| Design Choices for Multimodal Transformer Learning with Missing and Longitudinal Healthcare Data This paper studies transformer architectural choices for handling missing modalities and longitudinal observations in multimodal healthcare datasets. Ranked #471Preprint0/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Studies multimodal transformer architectures and fusion strategies across heterogeneous clinical data types. | |
| SkillForgeBench: Evaluating Agent Reliability Grounded in Agent Skill Specifications This paper introduces a benchmark that uses agent skill specifications to evaluate the reliability and procedural correctness of AI agents. Ranked #471New testPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:17 | On-topic Measures practical agent reliability and strict procedural compliance in deployment-like tasks. | |
| Unlocking the Regulatory Genome by ARGUS: An Evidence-Constrained Agentic Framework for Interpreting Single Nucleotide Variants Proposes ARGUS, an agentic framework separating deterministic genomic models from LLM reasoning to interpret noncoding single nucleotide variants reliably. Ranked #471New way to see a problemPreprint5/10 suggested | AI agents via arxiv · 12 Oct, 16:28 | On-topic Designs a multi-step investigation loop where an agent planner and verifier coordinate external tools. | |
| ReFormBench: Benchmarking the Stability of Response-Conditioned Format Correction in Large Language Models The paper presents a benchmark evaluating how reliably language models correct and maintain formatting requirements across changing instructions. Ranked #472New testPreprint5/10 suggested | LLM evaluation and benchmarks via arxiv · 15 Oct, 15:06 | On-topic Introduces ReFormBench to evaluate instruction compliance and format correction stability in LLMs. | |
| Axiom-0: Hierarchical Self-Refinement for Vision-Language-Action Policies This paper introduces a vision-language-action policy using hierarchical self-refinement across internal representations and action interfaces. Ranked #472Preprint0/10 suggested | Multimodal models via arxiv · 15 Oct, 06:18 | On-topic Develops a vision-language-action policy using hierarchical refinement across multimodal representations. | |
| Agent-Guided Evidence Acquisition for Emotion-Cause Pair Extraction This paper investigates an agent framework that selectively queries expert judgments for borderline emotion-cause pairs to improve extraction accuracy. Ranked #472Surprising resultPreprint5/10 suggested | AI agents via arxiv · 15 Oct, 06:18 | On-topic Explores agent-guided evidence acquisition where an agent decides when to query expert judgments. |
2,581–2,640 of 3,333← PreviousNext →