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 |
|---|---|---|---|
| EVA-Cap: Optimizing Audiovisual Video Captioning via Event-Centric Alignment Proposes an event-centric framework to align audio and video features when generating detailed video descriptions with language models. Ranked #386Preprint0/10 suggested | Multimodal models via arxiv · 17 Oct, 06:17 | On-topic Proposes an audiovisual captioning method aligning video and audio modalities. | |
| OntoAct: Hypothesis-Based Neuro-Symbolic Agents through Ontological Abstraction The paper pairs language models with symbolic verifiers to test and update reusable hypotheses for environment diagnostics and repair. Ranked #386New testPreprint6/10 suggested | LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17 | On-topic Introduces Tacit-FLE, a new benchmark for diagnosing and repairing hidden system faults. | |
| Online Tokenization with Optimal Delay in Linear Time The paper introduces an online tokenization algorithm that processes text streams in linear time with minimal algorithmic delay. Ranked #386Preprint0/10 suggested | Small and efficient models via arxiv · 15 Oct, 15:06 | On-topic Optimizes online tokenization algorithms to reduce latency and memory footprint in language models. | |
| Look Back to Move Forward: Learning to Revisit for Agentic Compositional Image Generation Develops an agentic text-to-image framework allowing agents to revisit and select earlier image states for iterative compositional generation. Ranked #386New way to see a problemPreprint6/10 suggested | AI agents via arxiv · 15 Oct, 06:17 | On-topic Proposes an agentic framework that plans iterative multi-step editing actions and selects states. | |
| Space Over Statistic, Retraining Over Ranking: Auditing Training Labels Under Label-Only Backdoors The paper evaluates label-only backdoor defenses and demonstrates that inspecting representations in unpoisoned feature spaces improves poison detection accuracy. Ranked #386Surprising resultPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 12 Oct, 16:28 | On-topic Addresses detecting and auditing label-only backdoor poisoning attacks in neural networks. | |
| AgentPerturb: Evaluating LLM Agent Robustness under Controlled Interaction Perturbations The authors introduce a framework to benchmark language model agent robustness under interaction perturbations across diverse tasks. Ranked #387New testPreprint6/10 suggested | LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17 | On-topic Proposes a benchmark framework to evaluate agent robustness under noise. | |
| DAGFormer: Global Topology Prediction for Transformer Cross-Layer Connections The paper proposes a Transformer architecture that plans dynamic cross-layer connections globally across the model based on input tokens. Ranked #387Preprint0/10 suggested | Small and efficient models via arxiv · 15 Oct, 15:06 | On-topic Introduces an efficient dynamic cross-layer connection architecture that improves performance under compact model budgets. | |
| 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 #387New way to see a problemPreprint6/10 suggested | AI agents via arxiv · 15 Oct, 06:17 | On-topic Develops a PID-gated mechanism for continual instruction adaptation in LLM agents. | |
| Heuristics are Not Enough: Calibrated and Corrected Caching for Video Diffusion Acceleration Accelerates video diffusion models by using closed-loop calibrated caching instead of fixed heuristic thresholds across denoising steps. Ranked #387Preprint0/10 suggested | Multimodal models via arxiv · 15 Oct, 06:17 | On-topic Focuses on accelerating video generation models via feature caching across diffusion steps. | |
| Not Every Change Is Necessary: Recoverable Drift in Large Language Model Unlearning The paper introduces a method to unlearn target knowledge while recovering collateral capability loss in non-target tasks. Ranked #387New way to see a problemPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 12 Oct, 16:28 | On-topic Develops targeted knowledge unlearning to remove unwanted behavior while maintaining model reliability. | |
| When Does Muon Actually Win? A Paired Data-, Time-, and Tuning-Budget Benchmark The paper compares the Muon optimizer against AdamW across controlled data, wall-clock time, and tuning budgets. Ranked #388Surprising resultPreprint6/10 suggested | LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17 | On-topic Provides a controlled benchmark comparing optimizers across data and time budgets. | |
| From Candidate Tokens to Final Outcomes: Decoding Relative Long-Horizon Candidate Value from Reasoning Model Representations The paper constructs a dataset to show that relative candidate token value in multi-step reasoning is linearly decodable from model representations. Ranked #388New way to see a problemPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:18 | On-topic Probes internal representations to predict reasoning failure and improve generation reliability. | |
| RIECNN: Real-Time Image Enhanced CNN for Traffic Sign Recognition Presents an enhanced convolutional neural network architecture for real-time traffic sign classification in autonomous driving applications. Ranked #388Peer-reviewed0/10 suggested | Small and efficient models via arxiv · 15 Oct, 06:17 | On-topic Designs a real-time CNN achieving low-latency sub-1.3ms inference for edge deployment. | |
| SkillFusion: Towards Reliable Multi-Skill Composition for LLM Agents This paper introduces an offline compiler that detects conflicts and merges multiple agent skills into a single coherent capability. Ranked #388New way to see a problemPreprint6/10 suggested | AI agents via arxiv · 15 Oct, 06:17 | On-topic Presents a compiler to reconcile conflicting skills for LLM agents. | |
| Sonic-AMG: Streaming Human Motion Generation from Spatial Audio Introduces a streaming causal transformer framework for generating human motion continuously from spatial audio inputs. Ranked #388Preprint0/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Integrates spatial audio signals with visual human motion generation. | |
| Can LLMs Tell Stories Out of Order? A Corpus and Benchmark for Nonlinear Narrative Generation Introduces a benchmark and evaluation to test whether language models can successfully generate coherent nonlinear narratives. Ranked #389New testPreprint6/10 suggested | LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17 | On-topic Introduces a new benchmark and evaluation methodology assessing nonlinear narrative reasoning in LLMs. | |
| Making LoRA Identifiable: Orthogonal Alignment for Continual Learning The paper introduces an orthogonal parameter alignment approach for low-rank adaptation to mitigate catastrophic forgetting in continual learning. Ranked #389Preprint0/10 suggested | Small and efficient models via arxiv · 15 Oct, 06:18 | On-topic Improves continual LoRA training efficiency and prevents memory scaling with tasks. | |
| Do correct answers have stable latent knowledge support? The paper proposes stability metrics to measure whether correct outputs in language models are backed by stable internal knowledge representations. Ranked #389New testPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:18 | On-topic Evaluates whether internal knowledge support is stable across elicitations to ensure reliability. | |
| 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 #389New testPreprint6/10 suggested | AI agents via arxiv · 15 Oct, 06:17 | On-topic Analyzes how LLM agents handle tool-call retries and timeout failures. | |
| X-AVDD: Cross-Attentive Audio-Visual Dataset Distillation Proposes a memory-efficient cross-attention method for distilling large paired audio-visual datasets into compact synthetic sets. Ranked #389Preprint0/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Directly investigates cross-attentive modeling to capture joint semantic correspondences across audio and video modalities. | |
| TurboMamba: Faster Inference, Faster Training, and Better Prediction with Lexical Memory Proposes a hybrid architecture combining global state-space modeling with local transformer attention and lexical memory for efficient inference. Ranked #390Preprint0/10 suggested | Small and efficient models via arxiv · 17 Oct, 06:17 | On-topic Proposes TurboMamba to achieve faster inference and training throughput with lower latency. | |
| ARGUS: An Agenda-Driven Framework for Reliable User Simulation The paper introduces an agenda-driven neuro-symbolic framework to improve the reliability of user simulation when testing interactive conversational agents. Ranked #390New way to see a problemPreprint6/10 suggested | LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17 | On-topic Introduces an agenda-driven evaluation framework and rubric-based metric for agents. | |
| Where Did We Go Wrong? Localizing Errors to Repair Multi-Turn Conversations The paper presents a framework to probabilistically localize user-intent misinterpretations and repair conversational states in multi-turn assistant systems. Ranked #390New way to see a problemPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:18 | On-topic Improves conversational assistant reliability by diagnosing and repairing misunderstood intent states. | |
| Google Earth AI Geospatial Reasoning Agent: Capabilities and a Systematic Evaluation The paper presents Google Earth AI's geospatial reasoning agent along with evaluation benchmarks for complex multi-step geospatial workflows. Ranked #390New testPreprint6/10 suggested | AI agents via arxiv · 15 Oct, 06:17 | On-topic Introduces a geospatial reasoning agent that decomposes queries and orchestrates tools. | |
| RARA: Robust Policy Fine-Tuning through Retrieval-Anchored Representation Alignment for Robotic Manipulation Aligns robot policy representations with retrieved pretraining examples to maintain generalization during low-data fine-tuning. Ranked #390Preprint0/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Adapts vision-language-action (VLA) multimodal models for downstream robotic manipulation tasks. | |
| LLM-Based Weak Annotation for Cybersecurity Tweet Classification: A Teacher-Student Framework The paper presents a teacher-student framework using language models for weak annotation to classify cybersecurity tweets. Ranked #391Peer-reviewed0/10 suggested | Small and efficient models via arxiv · 17 Oct, 06:17 | On-topic Distills LLM teacher knowledge into smaller domain-specific pre-trained language models. | |
| A Jury, Not an Electorate: Label-Free Reliability Estimation for Ensembles of Language Models Estimates language model ensemble member reliability from unlabeled agreement patterns using an expectation-maximization jury formulation. Ranked #391New way to see a problemPreprint6/10 suggested | LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17 | On-topic Presents a label-free reliability estimation and aggregation method for evaluating and ensembling LLMs. | |
| Social Bias in Chinese Large Language Models under Social Media Expressions This paper evaluates social bias in Chinese language models across social media expressions, internet slang, and sarcastic packaging. Ranked #391New testPeer-reviewed6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:18 | On-topic Examines social bias and alignment vulnerabilities in language models. | |
| SpecForge: Recursive Self-Improvement for Coding Agents via Behavioral Specifications The paper presents a self-improvement method where coding agents refine natural-language behavioral specifications using execution-verified tests. Ranked #391New way to see a problemPreprint6/10 suggested | AI agents via arxiv · 15 Oct, 06:17 | On-topic Develops a self-improvement search loop for autonomous coding agents. | |
| Destination-Conditioned Adaptation with Semantic-Geometric Alignment for Generative Inbetweening Adapts latent video diffusion models with target conditioning and semantic-geometric alignment for coherent generative frame inbetweening. Ranked #391Preprint0/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Focuses on generative image-to-video diffusion models and spatio-temporal feature alignment. | |
| 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 #392Preprint0/10 suggested | Small and efficient models via arxiv · 17 Oct, 06:17 | On-topic Applies distillation methods to build controlled, gender-agnostic speech models. | |
| OmicsBench: a long-horizon multi-turn benchmark of AI agents on research-study-scale bioinformatics Presents a multi-turn benchmark measuring language agent capabilities on complex, long-horizon bioinformatics workflows. Ranked #392New testPreprint6/10 suggested | LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17 | On-topic Introduces OmicsBench, a multi-turn benchmark for testing long-horizon agent capabilities. | |
| When Skills Remain But Access Fails: Factorized Skill Memory for Continual Skill Internalization This paper investigates catastrophic forgetting in continual agent skill learning and proposes a factorized memory architecture to preserve procedural execution. Ranked #392Surprising resultPreprint6/10 suggested | AI agents via arxiv · 15 Oct, 06:17 | On-topic Studies how language model agents retain and execute procedural skills continually. | |
| VeriPIE: Checkable Query Plans and Evidence-Guided Selection for Text-to-SQL The paper presents a neurosymbolic framework combining query plan checks and candidate selection to improve text-to-SQL translation for financial governance. Ranked #392New way to see a problemPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:17 | On-topic Provides symbolic validation and model risk management to ensure reliability of generated SQL queries. | |
| GuidedNav: Shaping Vision-Language Navigation Representations through Pre-action Attention and Spatial Guidance Guides vision-language navigation training using progress attention and spatial structure losses to improve action grounding and generalization. Ranked #392Preprint0/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Investigates cross-modal spatial and instruction grounding representations in vision-and-language navigation. | |
| PremiseCheck: Efficient Fact-Checking of Human and AI Peer Reviews Combines language models and lightweight decision models to fact-check statements in peer reviews against original manuscripts. Ranked #393New way to see a problemPreprint6/10 suggested | LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17 | On-topic Introduces ReviewFact, a new benchmark measuring fact-checking accuracy, cost, and latency. | |
| The Company a Question Keeps: Auditing Open Jev-Style Models for Retrieval This paper audits request independence in open Jev-style retrieval models across multi-passage query states. Ranked #393Surprising resultPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:18 | On-topic Examines behavioral inconsistency and input sensitivity risks in retrieval models. | |
| State Representation Shapes Long-Horizon Agent Learning and Transfer: An Empirical Study This empirical study evaluates how compact versus full state representations affect the ability of language model agents to learn long-horizon workflows. Ranked #393Surprising resultPreprint6/10 suggested | AI agents via arxiv · 15 Oct, 06:17 | On-topic Studies how state representations affect long-horizon action learning and tool workflows in agents. | |
| 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 #393Preprint0/10 suggested | Small and efficient models via arxiv · 15 Oct, 06:17 | On-topic Compresses and prunes image tokens to reduce computation and improve vision-language inference efficiency. | |
| Robust Vision-Language Navigation with Awareness of Instruction-Execution Progress Integrates execution progress awareness into vision-language navigation to suppress completed instruction segments and improve multi-step trajectory matching. Ranked #393Preprint0/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Addresses multimodal integration of visual observations and multi-step language instructions in navigation. | |
| What do AI chatbots do after users decline therapeutic redirection for suicide thoughts? Audits how persistent conversational agents remain with therapeutic crisis referrals after users repeatedly refuse help for suicidal thoughts. Ranked #394Surprising resultPreprint6/10 suggested | LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17 | On-topic Audits and measures conversational safety behaviors of commercial chatbots across standardized test conversations. | |
| Not All Challenges Are Equal: Evaluating Interactive Liveness Detection Against Real-Time Face Swaps The paper evaluates the effectiveness of randomized interactive gesture challenges in detecting real-time face-swap attacks during remote identity verification. Ranked #394New testPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:18 | On-topic Analyzes deepfake face-swap attack detection and verification robustness for deployment. | |
| Self-Evolving Visual Agent via Margin Reward This paper introduces a self-evolution framework that trains visual agents to invoke tools using reinforcement learning and unlabeled images. Ranked #394New way to see a problemPreprint6/10 suggested | AI agents via arxiv · 15 Oct, 06:17 | On-topic Presents a self-evolving visual agent that learns multi-step tool invocation. | |
| Kwai Summary Attention V2 Technical Report This technical report presents an updated summary-token attention architecture to optimize cache compression and computation across model layers. Ranked #394Preprint0/10 suggested | Small and efficient models via arxiv · 15 Oct, 06:17 | On-topic Compresses KV cache and reduces computation cost for efficient LLM serving and inference. | |
| AV-FlowEdit: Training-Free Prompt-Driven Audio-Visual Editing with Event-Aware Spectral Trajectory Preservation Presents a training-free framework for prompt-driven audio-video editing that uses flow differences to balance cross-modal changes and preservation. Ranked #394Preprint0/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Develops a framework for joint audio-visual cross-modal alignment and diffusion-based multimodal editing. | |
| Entertain, then Cancel: How Thinking Makes Reasoning Models Read Some and Or Literally Examines how extended reasoning steps lead language models to interpret everyday logical terms strictly literally instead of inferring pragmatic meaning. Ranked #395Surprising resultPreprint6/10 suggested | LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17 | On-topic Examines evaluation pitfalls in how reasoning models interpret pragmatic language constructs. | |
| Adversarial Robustness of Camera-Radar Sensor Fusion for Autonomous Driving The paper evaluates the adversarial robustness of five camera-radar fusion algorithms against sensor perturbations in autonomous driving settings. Ranked #395Surprising resultPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:18 | On-topic Measures adversarial robustness and failure modes in camera-radar multi-sensor fusion systems. | |
| 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 #395Surprising resultPreprint6/10 suggested | AI agents via arxiv · 15 Oct, 06:17 | On-topic Investigates activation steering as an internal memory mechanism for persistent agent experiences. | |
| Train for Many, Update with One: Improving Multi-Sample Reasoning via Fenchel-Bregman Posterior Learning This paper introduces a posterior learning approach that improves multi-sample reasoning coverage by updating models using single posterior draws. Ranked #395Preprint0/10 suggested | Small and efficient models via arxiv · 15 Oct, 06:17 | On-topic Proposes an efficient training method that reduces compute and sample draws for multi-sample reasoning. | |
| UniMergeTrack: Hierarchical Model Merging for General RGB-X Visual Object Tracking Proposes a hierarchical model merging method to combine modality-specific trackers into a unified RGB-X tracking framework. Ranked #395Preprint0/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Studies multimodal tracking integrating complementary sensor modalities like thermal infrared and depth. | |
| AgentEval: Evidence-Bounded Assessment of Stateful Agents Presents an evaluation system that tracks state and evidence to ensure autonomous agents fulfill obligations across long-horizon executions. Ranked #396New testPreprint6/10 suggested | LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17 | On-topic Introduces a system and methodology for evaluating stateful agent actions and evidence. | |
| From Observations to Authorized Actions: Measuring Source Information in Android Agents The paper examines Android agent security by measuring source provenance gaps in WebView interfaces and testing monitors against unauthorized actions. Ranked #396New testPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:18 | On-topic Proposes authorization monitors to prevent mobile agents from executing untrusted, malicious external actions. | |
| 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 #396New testPreprint6/10 suggested | AI agents via arxiv · 15 Oct, 06:17 | On-topic Introduces a benchmark evaluating LLM agent reliability and execution across complex sandboxed workflows. | |
| Explore with Memory, Learn without It: In-Context Exploration and Self-Distillation for LLM Reasoning This paper presents a reinforcement learning approach that uses in-context dynamic memory during rollouts and self-distillation to accelerate reasoning exploration. Ranked #396Preprint0/10 suggested | Small and efficient models via arxiv · 15 Oct, 06:17 | On-topic Uses self-distillation and in-context memory to improve learning and compute efficiency. | |
| Redirecting Teacher Supervision: Temporal-Response-Guided Latent Distillation for Video-Language Models Proposes Temporal-Response Guidance to direct latent distillation supervision for video-language models using local temporal perturbation responses. Ranked #396Preprint0/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Improves video-language models through temporal-response distillation for multimodal video understanding. | |
| Open Benchmark of AI Impact on Humans (ImpactBench): an expert-guided, open platform for the holistic evaluation of AI impact on humans The paper introduces an expert-guided benchmark to measure the impact of generative language models on user wellbeing. Ranked #397New testPreprint6/10 suggested | LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17 | On-topic Introduces ImpactBench, an expert-guided benchmark measuring conversational model impacts across 809 metrics. | |
| Only What You've Signed up For: Secure Workflows with Constrained Diffusion Agents The paper presents a framework enforcing security policies during the denoising steps of diffusion language models to prevent prompt injection. Ranked #397New way to see a problemPreprint6/10 suggested | AI safety and reliability in deployment via arxiv · 15 Oct, 06:18 | On-topic Introduces policy-constrained diffusion to protect agent workflows against prompt-injection attacks during generation. | |
| WorkBorne: Discovering Agentic Tasks from Real-World Artifact Transitions This paper presents a framework for discovering and synthesizing terminal-based agent tasks grounded in real-world artifact transitions. Ranked #397New way to see a problemPreprint6/10 suggested | AI agents via arxiv · 15 Oct, 06:17 | On-topic Develops synthetic terminal environments and interactive tasks for long-horizon agent execution. | |
| ChampionKV: Head-Wise Adaptive Budget Allocation for KV Cache Compression ChampionKV adaptively allocates key-value cache memory budgets across attention heads using observation-window concentration proxies. Ranked #397Preprint0/10 suggested | Small and efficient models via arxiv · 15 Oct, 06:17 | On-topic Introduces input-adaptive KV cache compression to make long-context LLM inference more efficient. | |
| BLEND: Balancing Personalization vs. Generalization in Federated Vision-Language Models Introduces a federated vision-language model tuning framework combining global shared adapters with client-specific personalized projection adapters. Ranked #397Preprint0/10 suggested | Multimodal models via arxiv · 12 Oct, 16:28 | On-topic Develops a personalized fine-tuning framework specifically for vision-language models. |
2,221–2,280 of 3,333← PreviousNext →