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
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. Preprint	AI safety and reliability in deployment via arxiv · 17 Oct, 06:17	Not relevant Addresses image distribution drift in editing rather than AI safety or reliability guardrails.
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. Preprint	AI agents via arxiv · 17 Oct, 06:17	Not relevant Focuses on multi-turn image generation and editing rather than language model agents.
Designing Transformers for Continuous Diffusion Language Models The authors introduce LangDiT, an adapted transformer architecture designed specifically for continuous diffusion language models. Preprint	AI agents via arxiv · 17 Oct, 06:17	Not relevant Develops continuous diffusion language model architectures rather than autonomous planning agents.
Designing Transformers for Continuous Diffusion Language Models The authors introduce LangDiT, an adapted transformer architecture designed specifically for continuous diffusion language models. Preprint	AI safety and reliability in deployment via arxiv · 17 Oct, 06:17	Not relevant Focuses on diffusion model perplexity optimization rather than safety, red-teaming, or guardrails.
Designing Transformers for Continuous Diffusion Language Models The authors introduce LangDiT, an adapted transformer architecture designed specifically for continuous diffusion language models. Preprint	LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17	Not relevant Proposes a model architecture rather than an evaluation framework or benchmark.
KnowGPO: Knowledge-Grouped Policy Optimization for Search Agents The authors propose KnowGPO, a reinforcement learning framework that assigns credit to deep search agents based on acquired knowledge conditions rather than trajectories. Preprint	AI safety and reliability in deployment via arxiv · 17 Oct, 06:17	Not relevant Studies reward modeling for search agents rather than safety guardrails or reliability risks.
KnowGPO: Knowledge-Grouped Policy Optimization for Search Agents The authors propose KnowGPO, a reinforcement learning framework that assigns credit to deep search agents based on acquired knowledge conditions rather than trajectories. Preprint	LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17	Not relevant Focuses on training policy optimization for search agents rather than evaluation benchmarks.
Evaluation Awareness Beyond English: Do LLMs Act Differently When Tested in Indic Languages? The authors evaluate how language models alter their refusal behavior when informed they are being monitored in English and five Indic languages. Preprint	AI agents via arxiv · 17 Oct, 06:17	Not relevant Studies evaluation awareness in LLMs, not multi-step agent planning or tool execution.
Evaluation Awareness Beyond English: Do LLMs Act Differently When Tested in Indic Languages? The authors evaluate how language models alter their refusal behavior when informed they are being monitored in English and five Indic languages. Preprint	LLMs for code via arxiv · 17 Oct, 06:17	Not relevant Evaluates multilingual safety behavior, not code generation or software engineering.
Evaluation Awareness Beyond English: Do LLMs Act Differently When Tested in Indic Languages? The authors evaluate how language models alter their refusal behavior when informed they are being monitored in English and five Indic languages. Preprint	Retrieval-augmented generation (RAG) via arxiv · 17 Oct, 06:17	Not relevant Does not evaluate or develop retrieval-augmented generation systems.
Evaluation Awareness Beyond English: Do LLMs Act Differently When Tested in Indic Languages? The authors evaluate how language models alter their refusal behavior when informed they are being monitored in English and five Indic languages. Preprint	Multimodal models via arxiv · 17 Oct, 06:17	Not relevant Focuses on text evaluation across languages rather than multimodal vision-language models.
Evaluation Awareness Beyond English: Do LLMs Act Differently When Tested in Indic Languages? The authors evaluate how language models alter their refusal behavior when informed they are being monitored in English and five Indic languages. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Does not cover model compression, pruning, or efficient inference architectures.
Dexterity from Supervised Play The authors introduce DSP, a framework where robots learn from brief human play and autonomously collect contact-rich data for dexterous manipulation. Preprint	LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17	Not relevant Presents a robotic manipulation framework rather than an LLM evaluation benchmark.
Dexterity from Supervised Play The authors introduce DSP, a framework where robots learn from brief human play and autonomously collect contact-rich data for dexterous manipulation. Preprint	AI safety and reliability in deployment via arxiv · 17 Oct, 06:17	Not relevant Investigates robotic manipulation policies rather than deployment safety or hallucination.
Dexterity from Supervised Play The authors introduce DSP, a framework where robots learn from brief human play and autonomously collect contact-rich data for dexterous manipulation. Preprint	AI agents via arxiv · 17 Oct, 06:17	Not relevant Deals with robotic manipulation and model-predictive control, not language model agents.
Dexterity from Supervised Play The authors introduce DSP, a framework where robots learn from brief human play and autonomously collect contact-rich data for dexterous manipulation. Preprint	Computer vision for industrial applications via arxiv · 17 Oct, 06:17	Not relevant Focuses on general robotic manipulation rather than industrial visual defect inspection.
3DCodeVerse: Exploring 3D Worlds as Code The authors introduce 3DCodeVerse, a dataset and multi-agent framework enabling coding agents to generate, reason over, and verify 3D executable code. Preprint	Computer vision for industrial applications via arxiv · 17 Oct, 06:17	Not relevant Focuses on 3D code generation, not industrial manufacturing defect detection.
3DCodeVerse: Exploring 3D Worlds as Code The authors introduce 3DCodeVerse, a dataset and multi-agent framework enabling coding agents to generate, reason over, and verify 3D executable code. Preprint	LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17	Not relevant Introduces a dataset and harness for 3D code rather than LLM benchmark evaluation methodology.
3DCodeVerse: Exploring 3D Worlds as Code The authors introduce 3DCodeVerse, a dataset and multi-agent framework enabling coding agents to generate, reason over, and verify 3D executable code. Preprint	AI safety and reliability in deployment via arxiv · 17 Oct, 06:17	Not relevant Focuses on 3D asset generation rather than AI safety, red-teaming, or guardrails.
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. Preprint	AI agents via arxiv · 17 Oct, 06:17	Not relevant Develops 3D visual world representations rather than LLM-based autonomous agents.
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. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Does not investigate model quantization, pruning, or efficient inference.
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. Preprint	LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17	Not relevant Proposes a predictive 3D model rather than a benchmark or evaluation method for LLMs.
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. Preprint	AI safety and reliability in deployment via arxiv · 17 Oct, 06:17	Not relevant Focuses on view synthesis and 3D modeling rather than AI safety or guardrails.
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. Preprint	Computer vision for industrial applications via arxiv · 17 Oct, 06:17	Not relevant Focuses on autonomous driving 3D datasets, not industrial inspection or manufacturing quality control.
MLExpert: Towards a Generalist Agentic Model for Autonomous Machine Learning Engineering The authors propose a post-training method enabling autonomous language model agents to organize experiments and carry out machine learning engineering tasks. Preprint	Multimodal models via arxiv · 17 Oct, 06:17	Not relevant Focuses on code and text-based ML engineering rather than multimodal perception or reasoning.
MLExpert: Towards a Generalist Agentic Model for Autonomous Machine Learning Engineering The authors propose a post-training method enabling autonomous language model agents to organize experiments and carry out machine learning engineering tasks. Preprint	Retrieval-augmented generation (RAG) via arxiv · 17 Oct, 06:17	Not relevant Addresses autonomous ML engineering rather than retrieval-augmented generation techniques.
MLExpert: Towards a Generalist Agentic Model for Autonomous Machine Learning Engineering The authors propose a post-training method enabling autonomous language model agents to organize experiments and carry out machine learning engineering tasks. Preprint	LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17	Not relevant Presents a model and training method for ML engineering rather than an LLM benchmark.
MLExpert: Towards a Generalist Agentic Model for Autonomous Machine Learning Engineering The authors propose a post-training method enabling autonomous language model agents to organize experiments and carry out machine learning engineering tasks. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Trains a 35B model for task automation rather than focusing on compression or efficient serving.
MLExpert: Towards a Generalist Agentic Model for Autonomous Machine Learning Engineering The authors propose a post-training method enabling autonomous language model agents to organize experiments and carry out machine learning engineering tasks. Preprint	AI safety and reliability in deployment via arxiv · 17 Oct, 06:17	Not relevant Investigates agent training for engineering tasks, not AI safety or reliability guardrails.
FDE-Gym: The Last Mile of AI Delivery The authors introduce FDE-Gym, a benchmark of 95 customer delivery tasks evaluating AI agents on ambiguous requirements and live software environments. Preprint	AI safety and reliability in deployment via arxiv · 17 Oct, 06:17	Not relevant Benchmarking delivery performance rather than safety guardrails, jailbreaks, or hallucination prevention.
FDE-Gym: The Last Mile of AI Delivery The authors introduce FDE-Gym, a benchmark of 95 customer delivery tasks evaluating AI agents on ambiguous requirements and live software environments. Preprint	Multimodal models via arxiv · 17 Oct, 06:17	Not relevant Evaluates software engineering agents without focusing on multimodal reasoning or vision-language models.
FDE-Gym: The Last Mile of AI Delivery The authors introduce FDE-Gym, a benchmark of 95 customer delivery tasks evaluating AI agents on ambiguous requirements and live software environments. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Does not explore model compression, quantization, or efficient inference methods.
FDE-Gym: The Last Mile of AI Delivery The authors introduce FDE-Gym, a benchmark of 95 customer delivery tasks evaluating AI agents on ambiguous requirements and live software environments. Preprint	Retrieval-augmented generation (RAG) via arxiv · 17 Oct, 06:17	Not relevant Focuses on customer delivery agent evaluation rather than retrieval-augmented generation architectures.
Rethinking Routing Exploration in Modern Mixture-of-Experts The authors propose MirrorMoE to reconcile an exploration dilemma between fine-grained expert segmentation and shared experts in mixture-of-experts models. Preprint	AI safety and reliability in deployment via arxiv · 17 Oct, 06:17	Not relevant Focuses on MoE parameter routing exploration rather than AI safety or reliability.
Rethinking Routing Exploration in Modern Mixture-of-Experts The authors propose MirrorMoE to reconcile an exploration dilemma between fine-grained expert segmentation and shared experts in mixture-of-experts models. Preprint	AI agents via arxiv · 17 Oct, 06:17	Not relevant Studies internal MoE routing dynamics rather than language model agent systems.
Rethinking Routing Exploration in Modern Mixture-of-Experts The authors propose MirrorMoE to reconcile an exploration dilemma between fine-grained expert segmentation and shared experts in mixture-of-experts models. Preprint	LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17	Not relevant Focuses on mixture-of-experts architectural design rather than LLM evaluation methodologies.
CoDA: Co-Evolution of Data Engines and Action Policies The authors present CoDA, a framework where a simulation data engine and a robot action policy co-evolve using memory of past failures. Preprint	AI agents via arxiv · 17 Oct, 06:17	Not relevant Addresses simulation data engines for robotic control, not LLM agent workflows.
CoDA: Co-Evolution of Data Engines and Action Policies The authors present CoDA, a framework where a simulation data engine and a robot action policy co-evolve using memory of past failures. Preprint	LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17	Not relevant Proposes policy training techniques for robots rather than LLM evaluation or benchmarks.
CoDA: Co-Evolution of Data Engines and Action Policies The authors present CoDA, a framework where a simulation data engine and a robot action policy co-evolve using memory of past failures. Preprint	Computer vision for industrial applications via arxiv · 17 Oct, 06:17	Not relevant Focuses on robotics policy learning rather than industrial defect detection or visual quality control.
CoDA: Co-Evolution of Data Engines and Action Policies The authors present CoDA, a framework where a simulation data engine and a robot action policy co-evolve using memory of past failures. Preprint	AI safety and reliability in deployment via arxiv · 17 Oct, 06:17	Not relevant Deals with simulation training for robots, not AI safety or deployment guardrails.
T2-z: Test-Time Optimization of Environment Latents for Robotic World Models The authors propose T2-Z, a method optimizing learnable environment latents at test time to help robotic world models adapt to unobserved physical factors. Preprint	Computer vision for industrial applications via arxiv · 17 Oct, 06:17	Not relevant Addresses test-time latents for robotics, not industrial manufacturing visual inspection.
T2-z: Test-Time Optimization of Environment Latents for Robotic World Models The authors propose T2-Z, a method optimizing learnable environment latents at test time to help robotic world models adapt to unobserved physical factors. Preprint	AI agents via arxiv · 17 Oct, 06:17	Not relevant Focuses on physical dynamics for robotic manipulation, not language-model agents.
T2-z: Test-Time Optimization of Environment Latents for Robotic World Models The authors propose T2-Z, a method optimizing learnable environment latents at test time to help robotic world models adapt to unobserved physical factors. Preprint	AI safety and reliability in deployment via arxiv · 17 Oct, 06:17	Not relevant Covers robotic world modeling rather than deployment safety, guardrails, or red-teaming.
T2-z: Test-Time Optimization of Environment Latents for Robotic World Models The authors propose T2-Z, a method optimizing learnable environment latents at test time to help robotic world models adapt to unobserved physical factors. Preprint	Multimodal models via arxiv · 17 Oct, 06:17	Not relevant Focuses on robotic dynamics optimization rather than multimodal foundation models or cross-modal reasoning.
T2-z: Test-Time Optimization of Environment Latents for Robotic World Models The authors propose T2-Z, a method optimizing learnable environment latents at test time to help robotic world models adapt to unobserved physical factors. Preprint	LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17	Not relevant Proposes a robotics world model rather than LLM benchmarking or evaluation methods.
Discovering state-of-the-art quantum circuits with large language models The authors introduce an evolutionary framework using large language models to design and optimize resource-efficient quantum circuits. Preprint	AI agents via arxiv · 17 Oct, 06:17	Not relevant Focuses on quantum circuit optimization using LLMs, not language model agent systems or planning frameworks.
Discovering state-of-the-art quantum circuits with large language models The authors introduce an evolutionary framework using large language models to design and optimize resource-efficient quantum circuits. Preprint	AI safety and reliability in deployment via arxiv · 17 Oct, 06:17	Not relevant Studies quantum algorithm discovery rather than AI safety, hallucination, or deployment reliability.
Discovering state-of-the-art quantum circuits with large language models The authors introduce an evolutionary framework using large language models to design and optimize resource-efficient quantum circuits. Preprint	LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17	Not relevant Deals with quantum circuit synthesis rather than evaluating or benchmarking language models.
LIBERO-PeRM: Benchmarking Personalized Robotic Manipulation The authors introduce LIBERO-PeRM, a simulation benchmark for studying how embodied robotic agents learn and follow user preferences in manipulation tasks. Preprint	LLM evaluation and benchmarks via arxiv · 17 Oct, 06:17	Not relevant Introduces a benchmark for embodied robotics rather than LLMs or language-model-based systems.
LIBERO-PeRM: Benchmarking Personalized Robotic Manipulation The authors introduce LIBERO-PeRM, a simulation benchmark for studying how embodied robotic agents learn and follow user preferences in manipulation tasks. Preprint	Computer vision for industrial applications via arxiv · 17 Oct, 06:17	Not relevant Evaluates household robotic manipulation, not industrial inspection or manufacturing visual quality control.
LIBERO-PeRM: Benchmarking Personalized Robotic Manipulation The authors introduce LIBERO-PeRM, a simulation benchmark for studying how embodied robotic agents learn and follow user preferences in manipulation tasks. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Does not investigate model compression, distillation, quantization, or efficient serving.
LIBERO-PeRM: Benchmarking Personalized Robotic Manipulation The authors introduce LIBERO-PeRM, a simulation benchmark for studying how embodied robotic agents learn and follow user preferences in manipulation tasks. Preprint	AI safety and reliability in deployment via arxiv · 17 Oct, 06:17	Not relevant Addresses preference satisfaction in robotics rather than AI safety, guardrails, or hallucination.
LIBERO-PeRM: Benchmarking Personalized Robotic Manipulation The authors introduce LIBERO-PeRM, a simulation benchmark for studying how embodied robotic agents learn and follow user preferences in manipulation tasks. Preprint	AI agents via arxiv · 17 Oct, 06:17	Not relevant Focuses on general robotics manipulation rather than LLM-based agent systems.
Reliability without Validity: A Systematic, Large-Scale Evaluation of LLM-as-a-Judge Models Across Agreement, Consistency, and Bias The paper evaluates twenty-one language models used as judges across three benchmarks, finding widespread discrepancies between agreement metrics and underlying validity. PreprintOlder paper (2026) Older paper (2026) — discussed this week	AI safety and reliability in deployment via blog · 16 Oct, 06:19	Not relevant Assesses general benchmark judge metrics rather than deployment safety mechanisms.
Reliability without Validity: A Systematic, Large-Scale Evaluation of LLM-as-a-Judge Models Across Agreement, Consistency, and Bias The paper evaluates twenty-one language models used as judges across three benchmarks, finding widespread discrepancies between agreement metrics and underlying validity. PreprintOlder paper (2026) Older paper (2026) — discussed this week	Multimodal models via blog · 16 Oct, 06:19	Not relevant Focuses on text evaluation benchmarks, not multimodal perception models.
Reliability without Validity: A Systematic, Large-Scale Evaluation of LLM-as-a-Judge Models Across Agreement, Consistency, and Bias The paper evaluates twenty-one language models used as judges across three benchmarks, finding widespread discrepancies between agreement metrics and underlying validity. PreprintOlder paper (2026) Older paper (2026) — discussed this week	LLMs for code via blog · 16 Oct, 06:19	Not relevant Does not study software engineering, code generation, or repair.
Reliability without Validity: A Systematic, Large-Scale Evaluation of LLM-as-a-Judge Models Across Agreement, Consistency, and Bias The paper evaluates twenty-one language models used as judges across three benchmarks, finding widespread discrepancies between agreement metrics and underlying validity. PreprintOlder paper (2026) Older paper (2026) — discussed this week	Retrieval-augmented generation (RAG) via blog · 16 Oct, 06:19	Not relevant Does not study retrieval-augmented generation architectures or document grounding.
Reliability without Validity: A Systematic, Large-Scale Evaluation of LLM-as-a-Judge Models Across Agreement, Consistency, and Bias The paper evaluates twenty-one language models used as judges across three benchmarks, finding widespread discrepancies between agreement metrics and underlying validity. PreprintOlder paper (2026) Older paper (2026) — discussed this week	AI agents via blog · 16 Oct, 06:19	Not relevant Evaluates LLM judges, not agent planning or tool use.
Reliability without Validity: A Systematic, Large-Scale Evaluation of LLM-as-a-Judge Models Across Agreement, Consistency, and Bias The paper evaluates twenty-one language models used as judges across three benchmarks, finding widespread discrepancies between agreement metrics and underlying validity. PreprintOlder paper (2026) Older paper (2026) — discussed this week	Small and efficient models via blog · 16 Oct, 06:19	Not relevant Focuses on evaluation methodology rather than model efficiency or compression.
Comprehensive Analysis and Experiment of Extreme Faults in MMC Based on IGCT The paper analyzes extreme faults in modular multilevel converters and proposes a protection scheme using integrated gate-commutated thyristors. Peer-reviewedOlder paper (2023) Older paper (2023) — discussed this week	AI safety and reliability in deployment via ieee · 16 Oct, 06:19	Not relevant Electrical engineering paper on power converters, not AI safety.

5,761–5,820 of 17,313← Previous Next →