Every paper checked — nothing hidden

All papers

Every paper checked for brief #4 of 12–19 Oct 2026 — nothing is hidden. Brief papers first.

All17,313 In brief24 More (on-topic)3,333 Not relevant13,754 Not assessed87 Older352 Could not confirm93 Blogs with no paper266 All blogs284

Paper	Topic	Relevance
Change the Number, Change the Answer: Assessing Rating Scale Sensitivity in LLM-Based Text Scoring The paper evaluates how choices in rating scale design systematically alter text evaluation scores produced by large language models. Preprint	Retrieval-augmented generation (RAG) via arxiv · 17 Oct, 06:17	Not relevant Examines rating scale sensitivity, not retrieval-augmented generation.
Change the Number, Change the Answer: Assessing Rating Scale Sensitivity in LLM-Based Text Scoring The paper evaluates how choices in rating scale design systematically alter text evaluation scores produced by large language models. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Investigates text scoring variability rather than model efficiency.
Change the Number, Change the Answer: Assessing Rating Scale Sensitivity in LLM-Based Text Scoring The paper evaluates how choices in rating scale design systematically alter text evaluation scores produced by large language models. Preprint	Multimodal models via arxiv · 17 Oct, 06:17	Not relevant Focuses on text scoring rather than multimodal inputs.
Change the Number, Change the Answer: Assessing Rating Scale Sensitivity in LLM-Based Text Scoring The paper evaluates how choices in rating scale design systematically alter text evaluation scores produced by large language models. Preprint	LLMs for code via arxiv · 17 Oct, 06:17	Not relevant Evaluates text scoring scales, not code generation or repair.
Terminal-RSI: Recursive Self-Improving Terminal Agents via Recursive Self-Rewrite and Policy-Refine The paper proposes an iterative self-improvement method combining trajectory rewriting and policy refinement for terminal-based coding agents. Preprint	AI safety and reliability in deployment via arxiv · 17 Oct, 06:17	Not relevant Focuses on terminal task optimization, not safety or guardrails.
Terminal-RSI: Recursive Self-Improving Terminal Agents via Recursive Self-Rewrite and Policy-Refine The paper proposes an iterative self-improvement method combining trajectory rewriting and policy refinement for terminal-based coding agents. Preprint	LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17	Not relevant Focuses on agent training methods rather than evaluation benchmark design.
AutoSFT: An Agentic System for Automated Data Research in Supervised Fine-Tuning The paper introduces a multi-agent system to automate data curation and dataset editing for supervised fine-tuning. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Studies data engineering workflows, not model compression or efficiency techniques.
AutoSFT: An Agentic System for Automated Data Research in Supervised Fine-Tuning The paper introduces a multi-agent system to automate data curation and dataset editing for supervised fine-tuning. Preprint	LLMs for code via arxiv · 17 Oct, 06:17	Not relevant Focuses on supervised fine-tuning data curation, not coding assistants.
AutoSFT: An Agentic System for Automated Data Research in Supervised Fine-Tuning The paper introduces a multi-agent system to automate data curation and dataset editing for supervised fine-tuning. Preprint	AI safety and reliability in deployment via arxiv · 17 Oct, 06:17	Not relevant Investigates fine-tuning automation, not deployment safety or guardrails.
AutoSFT: An Agentic System for Automated Data Research in Supervised Fine-Tuning The paper introduces a multi-agent system to automate data curation and dataset editing for supervised fine-tuning. Preprint	Retrieval-augmented generation (RAG) via arxiv · 17 Oct, 06:17	Not relevant Focuses on data curation for fine-tuning rather than RAG systems.
AutoSFT: An Agentic System for Automated Data Research in Supervised Fine-Tuning The paper introduces a multi-agent system to automate data curation and dataset editing for supervised fine-tuning. Preprint	Multimodal models via arxiv · 17 Oct, 06:17	Not relevant Deals with text data preparation rather than multimodal models.
AutoSFT: An Agentic System for Automated Data Research in Supervised Fine-Tuning The paper introduces a multi-agent system to automate data curation and dataset editing for supervised fine-tuning. Preprint	LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17	Not relevant Develops a fine-tuning agent system rather than an evaluation benchmark.
HTN-Orchestrate: A Hierarchical Task Network Framework for Multi-Agent Orchestration The paper introduces a hierarchical task network framework that separates multi-agent planning from execution scheduling. Preprint	Multimodal models via arxiv · 17 Oct, 06:17	Not relevant Does not involve multimodal models or vision-language processing.
HTN-Orchestrate: A Hierarchical Task Network Framework for Multi-Agent Orchestration The paper introduces a hierarchical task network framework that separates multi-agent planning from execution scheduling. Preprint	AI safety and reliability in deployment via arxiv · 17 Oct, 06:17	Not relevant Focuses on workflow efficiency, not safety, guardrails, or alignment.
HTN-Orchestrate: A Hierarchical Task Network Framework for Multi-Agent Orchestration The paper introduces a hierarchical task network framework that separates multi-agent planning from execution scheduling. Preprint	Retrieval-augmented generation (RAG) via arxiv · 17 Oct, 06:17	Not relevant Focuses on task planning orchestration rather than RAG or retrieval pipelines.
HTN-Orchestrate: A Hierarchical Task Network Framework for Multi-Agent Orchestration The paper introduces a hierarchical task network framework that separates multi-agent planning from execution scheduling. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Addresses multi-agent scheduling rather than model compression or inference efficiency.
HTN-Orchestrate: A Hierarchical Task Network Framework for Multi-Agent Orchestration The paper introduces a hierarchical task network framework that separates multi-agent planning from execution scheduling. Preprint	LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17	Not relevant Proposes an orchestration architecture, not new LLM evaluation benchmarks.
HTN-Orchestrate: A Hierarchical Task Network Framework for Multi-Agent Orchestration The paper introduces a hierarchical task network framework that separates multi-agent planning from execution scheduling. Preprint	LLMs for code via arxiv · 17 Oct, 06:17	Not relevant Evaluates multi-agent planning frameworks rather than code generation.
Decentralized LLM Workflows via Sequential Auctions The paper examines decentralized multi-agent coordination through sequential market auctions under conditions of imperfect competition. Preprint	AI safety and reliability in deployment via arxiv · 17 Oct, 06:17	Not relevant Deals with market mechanisms and bidding rather than safety or guardrails.
Decentralized LLM Workflows via Sequential Auctions The paper examines decentralized multi-agent coordination through sequential market auctions under conditions of imperfect competition. Preprint	LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17	Not relevant Focuses on multi-agent game-theoretic workflows rather than LLM evaluation benchmarks.
Beyond Downstream QA: Rethinking How We Evaluate Agent Memory The paper presents a benchmark that isolates agent memory evolution from downstream retrieval and generation failures. Preprint	Multimodal models via arxiv · 17 Oct, 06:17	Not relevant Evaluates text-based agent memory rather than multimodal reasoning.
Beyond Downstream QA: Rethinking How We Evaluate Agent Memory The paper presents a benchmark that isolates agent memory evolution from downstream retrieval and generation failures. Preprint	LLMs for code via arxiv · 17 Oct, 06:17	Not relevant Studies agent memory benchmarks, not software engineering or code generation.
Beyond Downstream QA: Rethinking How We Evaluate Agent Memory The paper presents a benchmark that isolates agent memory evolution from downstream retrieval and generation failures. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Focuses on memory evaluation, not model compression or efficient serving.
TRACE: Topology-Aware Reconnection via Prediction-Adaptive Connectivity Enforcement for Thin-Structure Segmentation The paper introduces an adaptive connectivity objective to improve deep learning segmentation of thin, continuous anatomical or physical structures. Preprint	Computer vision for industrial applications via arxiv · 17 Oct, 06:17	Not relevant Generic thin-structure segmentation, not industrial defect detection or visual quality control.
TRACE: Topology-Aware Reconnection via Prediction-Adaptive Connectivity Enforcement for Thin-Structure Segmentation The paper introduces an adaptive connectivity objective to improve deep learning segmentation of thin, continuous anatomical or physical structures. Preprint	AI safety and reliability in deployment via arxiv · 17 Oct, 06:17	Not relevant Addresses segmentation topology, not AI safety, jailbreaks, or guardrails.
TRACE: Topology-Aware Reconnection via Prediction-Adaptive Connectivity Enforcement for Thin-Structure Segmentation The paper introduces an adaptive connectivity objective to improve deep learning segmentation of thin, continuous anatomical or physical structures. Preprint	Multimodal models via arxiv · 17 Oct, 06:17	Not relevant Pure computer vision method without multimodal or language integration.
TRACE: Topology-Aware Reconnection via Prediction-Adaptive Connectivity Enforcement for Thin-Structure Segmentation The paper introduces an adaptive connectivity objective to improve deep learning segmentation of thin, continuous anatomical or physical structures. Preprint	AI agents via arxiv · 17 Oct, 06:17	Not relevant Focuses on image segmentation losses, not AI agents.
TRACE: Topology-Aware Reconnection via Prediction-Adaptive Connectivity Enforcement for Thin-Structure Segmentation The paper introduces an adaptive connectivity objective to improve deep learning segmentation of thin, continuous anatomical or physical structures. Preprint	LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17	Not relevant Evaluates computer vision segmentation, not LLM benchmarks or evaluation.
GNN-Guided Local Refinement for Efficient Layout Decomposition with DSA-MP The paper proposes a graph neural network method to speed up layout decomposition for advanced semiconductor manufacturing. Preprint	AI agents via arxiv · 17 Oct, 06:17	Not relevant Applies graph neural networks to lithography layout decomposition, not language agents.
GNN-Guided Local Refinement for Efficient Layout Decomposition with DSA-MP The paper proposes a graph neural network method to speed up layout decomposition for advanced semiconductor manufacturing. Preprint	AI safety and reliability in deployment via arxiv · 17 Oct, 06:17	Not relevant Focuses on semiconductor manufacturing layout algorithms rather than AI deployment safety.
GNN-Guided Local Refinement for Efficient Layout Decomposition with DSA-MP The paper proposes a graph neural network method to speed up layout decomposition for advanced semiconductor manufacturing. Preprint	LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17	Not relevant Focuses on circuit layout optimization rather than language model evaluation benchmarks.
Skill-Decoupled On-Policy Expert Distillation for Unified Vision-Language Models The paper introduces a skill-decoupled on-policy distillation method to merge vision-language experts without conflicting gradients or reward hacking. Preprint	LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17	Not relevant Evaluates a distillation algorithm rather than proposing a new evaluation benchmark.
Skill-Decoupled On-Policy Expert Distillation for Unified Vision-Language Models The paper introduces a skill-decoupled on-policy distillation method to merge vision-language experts without conflicting gradients or reward hacking. Preprint	Computer vision for industrial applications via arxiv · 17 Oct, 06:17	Not relevant Addresses general vision-language capabilities rather than industrial defect inspection.
Skill-Decoupled On-Policy Expert Distillation for Unified Vision-Language Models The paper introduces a skill-decoupled on-policy distillation method to merge vision-language experts without conflicting gradients or reward hacking. Preprint	AI agents via arxiv · 17 Oct, 06:17	Not relevant Focuses on multimodal post-training distillation, not agentic planning.
Skill-Decoupled On-Policy Expert Distillation for Unified Vision-Language Models The paper introduces a skill-decoupled on-policy distillation method to merge vision-language experts without conflicting gradients or reward hacking. Preprint	AI safety and reliability in deployment via arxiv · 17 Oct, 06:17	Not relevant Focuses on visual capability distillation without addressing safety risks or guardrails.
OS-Helix: Unifying GUI Interaction and Code Execution in Native Computer-Using Agents The paper presents an end-to-end trained multimodal agent that combines visual graphical interface actions and executable code execution. Preprint	Computer vision for industrial applications via arxiv · 17 Oct, 06:17	Not relevant Focuses on desktop GUI automation, not industrial visual inspection.
OS-Helix: Unifying GUI Interaction and Code Execution in Native Computer-Using Agents The paper presents an end-to-end trained multimodal agent that combines visual graphical interface actions and executable code execution. Preprint	AI safety and reliability in deployment via arxiv · 17 Oct, 06:17	Not relevant Focuses on agent task execution capability rather than safety or robustness guardrails.
OS-Helix: Unifying GUI Interaction and Code Execution in Native Computer-Using Agents The paper presents an end-to-end trained multimodal agent that combines visual graphical interface actions and executable code execution. Preprint	LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17	Not relevant Presents an agent training method rather than an evaluation benchmark.
OS-Helix: Unifying GUI Interaction and Code Execution in Native Computer-Using Agents The paper presents an end-to-end trained multimodal agent that combines visual graphical interface actions and executable code execution. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Does not address model pruning, quantization, or efficient inference serving.
Training-Free On-Policy Distillation The paper presents a closed-form, training-free approach to merge specialized fine-tuned models into a single student using forward passes. Preprint	LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17	Not relevant Proposes a training-free distillation method rather than an evaluation benchmark.
Training-Free On-Policy Distillation The paper presents a closed-form, training-free approach to merge specialized fine-tuned models into a single student using forward passes. Preprint	Multimodal models via arxiv · 17 Oct, 06:17	Not relevant Evaluates text language models without examining multimodal data.
Training-Free On-Policy Distillation The paper presents a closed-form, training-free approach to merge specialized fine-tuned models into a single student using forward passes. Preprint	LLMs for code via arxiv · 17 Oct, 06:17	Not relevant Presents a general distillation algorithm, not specific software engineering LLMs.
Training-Free On-Policy Distillation The paper presents a closed-form, training-free approach to merge specialized fine-tuned models into a single student using forward passes. Preprint	AI agents via arxiv · 17 Oct, 06:17	Not relevant Investigates weight distillation techniques, not autonomous agentic systems.
Training-Free On-Policy Distillation The paper presents a closed-form, training-free approach to merge specialized fine-tuned models into a single student using forward passes. Preprint	Retrieval-augmented generation (RAG) via arxiv · 17 Oct, 06:17	Not relevant Addresses model weight merging without involving document retrieval.
Training-Free On-Policy Distillation The paper presents a closed-form, training-free approach to merge specialized fine-tuned models into a single student using forward passes. Preprint	AI safety and reliability in deployment via arxiv · 17 Oct, 06:17	Not relevant Focuses on capability distillation rather than safety or reliability guardrails.
Shortcut Sensitivity in Agent-Safety Evaluations The paper introduces a framework to detect spurious shortcuts in safety benchmarks used to evaluate tool-using AI agents. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Focuses on benchmark shortcuts rather than model efficiency or distillation.
Shortcut Sensitivity in Agent-Safety Evaluations The paper introduces a framework to detect spurious shortcuts in safety benchmarks used to evaluate tool-using AI agents. Preprint	AI agents via arxiv · 17 Oct, 06:17	Not relevant Focuses on safety guard evaluation artifacts, not agent workflows.
Language Is a Security Surface for Agent Guards The paper evaluates 19 safety guards across multiple non-English languages to measure how translation alters step-level safety decisions for agents. Preprint	LLMs for code via arxiv · 17 Oct, 06:17	Not relevant Focuses on multilingual agent security, not programming or code synthesis.
Language Is a Security Surface for Agent Guards The paper evaluates 19 safety guards across multiple non-English languages to measure how translation alters step-level safety decisions for agents. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Evaluates guard sensitivity rather than model compression or efficient serving.
Language Is a Security Surface for Agent Guards The paper evaluates 19 safety guards across multiple non-English languages to measure how translation alters step-level safety decisions for agents. Preprint	Retrieval-augmented generation (RAG) via arxiv · 17 Oct, 06:17	Not relevant Does not involve external retrieval or document augmentation.
Language Is a Security Surface for Agent Guards The paper evaluates 19 safety guards across multiple non-English languages to measure how translation alters step-level safety decisions for agents. Preprint	AI agents via arxiv · 17 Oct, 06:17	Not relevant Studies security guardrails across languages rather than agent planning or workflows.
Language Is a Security Surface for Agent Guards The paper evaluates 19 safety guards across multiple non-English languages to measure how translation alters step-level safety decisions for agents. Preprint	Multimodal models via arxiv · 17 Oct, 06:17	Not relevant Focuses on text translations, not vision or multimodal processing.
Disagreement-Guided Weak-to-Strong Routing The paper develops an unsupervised routing method to escalate tasks from weak language models to stronger models without ground-truth labels. Preprint	AI agents via arxiv · 17 Oct, 06:17	Not relevant Studies model routing policies rather than autonomous agent planning architectures.
Disagreement-Guided Weak-to-Strong Routing The paper develops an unsupervised routing method to escalate tasks from weak language models to stronger models without ground-truth labels. Preprint	LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17	Not relevant Presents a routing methodology rather than an evaluation benchmark.
Disagreement-Guided Weak-to-Strong Routing The paper develops an unsupervised routing method to escalate tasks from weak language models to stronger models without ground-truth labels. Preprint	AI safety and reliability in deployment via arxiv · 17 Oct, 06:17	Not relevant Focuses on routing optimization rather than safety, guardrails, or alignment.
X-RIC: An Agentic RAN-Intelligent Controller Benchmark for Enhancing Reinforcement Learning The paper introduces an agentic reasoning controller framework and an evaluation benchmark for optimizing open radio access networks. Preprint	AI safety and reliability in deployment via arxiv · 17 Oct, 06:17	Not relevant Addresses telecommunications control optimization rather than AI safety or guardrails.
Librarian Agents: Training LLMs to Curate Long-Context Workspaces that Anticipate Future Queries The paper trains language model agents to reorganize unstructured document collections into structured workspaces to speed up future multi-document queries. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Does not address model compression, pruning, or efficient inference.
Librarian Agents: Training LLMs to Curate Long-Context Workspaces that Anticipate Future Queries The paper trains language model agents to reorganize unstructured document collections into structured workspaces to speed up future multi-document queries. Preprint	AI safety and reliability in deployment via arxiv · 17 Oct, 06:17	Not relevant Focuses on task performance and curation rather than safety or guardrails.
Librarian Agents: Training LLMs to Curate Long-Context Workspaces that Anticipate Future Queries The paper trains language model agents to reorganize unstructured document collections into structured workspaces to speed up future multi-document queries. Preprint	Multimodal models via arxiv · 17 Oct, 06:17	Not relevant Handles unstructured text documents rather than multimodal image/video understanding.
Librarian Agents: Training LLMs to Curate Long-Context Workspaces that Anticipate Future Queries The paper trains language model agents to reorganize unstructured document collections into structured workspaces to speed up future multi-document queries. Preprint	LLMs for code via arxiv · 17 Oct, 06:17	Not relevant Focuses on document workspace structuring rather than code generation or repair.

2,161–2,220 of 13,754← Previous Next →