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
LLM2R: From Large Language Models to 3D Reassemblers The paper adapts a causal language model using compressed 3D geometric tokens to assemble multi-part objects in a single forward pass. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Focuses on 3D geometric assembly rather than safety or reliability.
Beyond Answer Voting: Fine-Grained Correction of LLM Reasoning through Graph Alignment across Sampled Traces The paper proposes a framework that uses reasoning graphs to locate claim-level disagreements across sampled traces and correct multi-step language model reasoning. Preprint	Small and efficient models via arxiv · 15 Oct, 06:17	Not relevant Focuses on graph-aligned reasoning correction rather than model compression.
Beyond Answer Voting: Fine-Grained Correction of LLM Reasoning through Graph Alignment across Sampled Traces The paper proposes a framework that uses reasoning graphs to locate claim-level disagreements across sampled traces and correct multi-step language model reasoning. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Proposes an inference-time verification method rather than benchmark evaluation.
Beyond Answer Voting: Fine-Grained Correction of LLM Reasoning through Graph Alignment across Sampled Traces The paper proposes a framework that uses reasoning graphs to locate claim-level disagreements across sampled traces and correct multi-step language model reasoning. Preprint	Retrieval-augmented generation (RAG) via arxiv · 15 Oct, 06:17	Not relevant Aligns reasoning graphs across sampled traces rather than retrieving external documents.
Beyond Answer Voting: Fine-Grained Correction of LLM Reasoning through Graph Alignment across Sampled Traces The paper proposes a framework that uses reasoning graphs to locate claim-level disagreements across sampled traces and correct multi-step language model reasoning. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Focuses on multi-trace answer verification rather than interactive agent workflows.
Beyond Answer Voting: Fine-Grained Correction of LLM Reasoning through Graph Alignment across Sampled Traces The paper proposes a framework that uses reasoning graphs to locate claim-level disagreements across sampled traces and correct multi-step language model reasoning. Preprint	Multimodal models via arxiv · 15 Oct, 06:17	Not relevant Addresses text-based reasoning without any image or multimodal components.
Beyond Answer Voting: Fine-Grained Correction of LLM Reasoning through Graph Alignment across Sampled Traces The paper proposes a framework that uses reasoning graphs to locate claim-level disagreements across sampled traces and correct multi-step language model reasoning. Preprint	LLMs for code via arxiv · 15 Oct, 06:17	Not relevant Deals with general reasoning verification rather than software engineering or code.
DPOison-VL: Clean Preference Poisoning in Vision-Language Direct Preference Optimization The paper introduces an imperceptible image perturbation attack that manipulates alignment in vision-language models during preference optimization without altering text annotations. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Develops an attack method rather than evaluation frameworks or general benchmarks.
DPOison-VL: Clean Preference Poisoning in Vision-Language Direct Preference Optimization The paper introduces an imperceptible image perturbation attack that manipulates alignment in vision-language models during preference optimization without altering text annotations. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Studies preference poisoning attacks rather than autonomous agent systems.
DPOison-VL: Clean Preference Poisoning in Vision-Language Direct Preference Optimization The paper introduces an imperceptible image perturbation attack that manipulates alignment in vision-language models during preference optimization without altering text annotations. Preprint	Computer vision for industrial applications via arxiv · 15 Oct, 06:17	Not relevant Focuses on vision-language model security rather than industrial defect inspection.
Jev-ToM: Evaluating Theory of Mind in Typed Decision Models The paper introduces a benchmark to evaluate how typed decision models handle theory-of-mind reasoning tasks compared to standard generative language models. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Tests theory of mind selection rather than multi-step tool-using agents.
Jev-ToM: Evaluating Theory of Mind in Typed Decision Models The paper introduces a benchmark to evaluate how typed decision models handle theory-of-mind reasoning tasks compared to standard generative language models. Preprint	Retrieval-augmented generation (RAG) via arxiv · 15 Oct, 06:17	Not relevant Studies theory-of-mind question selection rather than retrieval-augmented generation.
Jev-ToM: Evaluating Theory of Mind in Typed Decision Models The paper introduces a benchmark to evaluate how typed decision models handle theory-of-mind reasoning tasks compared to standard generative language models. Preprint	LLMs for code via arxiv · 15 Oct, 06:17	Not relevant Examines typed cognitive models rather than code generation or software repair.
Jev-ToM: Evaluating Theory of Mind in Typed Decision Models The paper introduces a benchmark to evaluate how typed decision models handle theory-of-mind reasoning tasks compared to standard generative language models. Preprint	Multimodal models via arxiv · 15 Oct, 06:17	Not relevant Focuses on English text Theory of Mind benchmarks without multimodal data.
Jev-ToM: Evaluating Theory of Mind in Typed Decision Models The paper introduces a benchmark to evaluate how typed decision models handle theory-of-mind reasoning tasks compared to standard generative language models. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Studies cognitive modeling and selection accuracy rather than safety guardrails.
Nudge the Judge: Uncovering Program Decision Hijacking in Real-World Jev Applications The paper presents an automated framework to evaluate how adversarial contexts can redirect program execution through valid structured judgments in language-model applications. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Introduces a vulnerability testing framework for specific apps rather than LLM benchmarks.
Nudge the Judge: Uncovering Program Decision Hijacking in Real-World Jev Applications The paper presents an automated framework to evaluate how adversarial contexts can redirect program execution through valid structured judgments in language-model applications. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Studies program decision hijacking in decision models rather than autonomous agents.
Does self-distillation improve reasoning? The paper evaluates whether self-distillation using privileged information improves reasoning capabilities or reflects overfitting to training tasks. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Analyzes training mechanics rather than safety, guardrails, or alignment risks.
Does self-distillation improve reasoning? The paper evaluates whether self-distillation using privileged information improves reasoning capabilities or reflects overfitting to training tasks. Preprint	Multimodal models via arxiv · 15 Oct, 06:17	Not relevant Focuses on text-based reasoning rather than multimodal image or video models.
Does self-distillation improve reasoning? The paper evaluates whether self-distillation using privileged information improves reasoning capabilities or reflects overfitting to training tasks. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Investigates self-distillation training methods rather than agent architectures or tool use.
Does self-distillation improve reasoning? The paper evaluates whether self-distillation using privileged information improves reasoning capabilities or reflects overfitting to training tasks. Preprint	Retrieval-augmented generation (RAG) via arxiv · 15 Oct, 06:17	Not relevant Focuses on reasoning self-distillation rather than external document retrieval.
Does self-distillation improve reasoning? The paper evaluates whether self-distillation using privileged information improves reasoning capabilities or reflects overfitting to training tasks. Preprint	LLMs for code via arxiv · 15 Oct, 06:17	Not relevant Evaluates reasoning methods across domains rather than focusing on software engineering.
Can an Agent Learn to Predict Its Own Failure? Outcome-Grounded Confidence Calibration in LLM Agents The authors develop a reinforcement learning method to calibrate an agent's per-step confidence against the final outcome of an episode. Preprint	Small and efficient models via arxiv · 15 Oct, 06:17	Not relevant Addresses confidence calibration rather than model compression or efficient serving.
Knowledge Graphs as Files: An Agent Harness Where Tool Creation and Tool Use Co-Evolve This paper proposes a system where language model agents write code to create and refine tools directly over raw knowledge-graph files. Preprint	Multimodal models via arxiv · 15 Oct, 06:17	Not relevant Involves text-rich knowledge graphs without vision, audio, or multimodal data.
Knowledge Graphs as Files: An Agent Harness Where Tool Creation and Tool Use Co-Evolve This paper proposes a system where language model agents write code to create and refine tools directly over raw knowledge-graph files. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Focuses on agent search performance rather than safety guardrails or reliability.
Knowledge Graphs as Files: An Agent Harness Where Tool Creation and Tool Use Co-Evolve This paper proposes a system where language model agents write code to create and refine tools directly over raw knowledge-graph files. Preprint	Small and efficient models via arxiv · 15 Oct, 06:17	Not relevant Focuses on tool creation architectures rather than model compression or efficiency.
Knowledge Graphs as Files: An Agent Harness Where Tool Creation and Tool Use Co-Evolve This paper proposes a system where language model agents write code to create and refine tools directly over raw knowledge-graph files. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Evaluates an agent system on benchmarks rather than introducing evaluation frameworks.
Two Routes to Interoceptive Prioritisation: Dissociating Likelihood Precision and Preference Reweighting in Active Inference The authors compare information precision and preference reweighting mechanisms in active-inference agents managing competing survival needs. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Active inference neuroscience study on biological agents, not LLM agent systems.
Two Routes to Interoceptive Prioritisation: Dissociating Likelihood Precision and Preference Reweighting in Active Inference The authors compare information precision and preference reweighting mechanisms in active-inference agents managing competing survival needs. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Investigates biological active inference models rather than AI safety or reliability.
Two Routes to Interoceptive Prioritisation: Dissociating Likelihood Precision and Preference Reweighting in Active Inference The authors compare information precision and preference reweighting mechanisms in active-inference agents managing competing survival needs. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Studies active inference simulations rather than LLM benchmarks or evaluation.
Falsification-proof non-market allocation mechanisms The paper characterizes optimal falsification-proof non-market allocation mechanisms when individuals can strategically manipulate their observable characteristics. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Theoretical economic study with no connection to LLM benchmarks or evaluation.
Falsification-proof non-market allocation mechanisms The paper characterizes optimal falsification-proof non-market allocation mechanisms when individuals can strategically manipulate their observable characteristics. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Theoretical economics paper on allocation mechanisms, not language model agents.
Falsification-proof non-market allocation mechanisms The paper characterizes optimal falsification-proof non-market allocation mechanisms when individuals can strategically manipulate their observable characteristics. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Deals with economic resource allocation rather than AI deployment safety.
Geometric Calibration of Supervised Fine-Tuning This paper presents a post-hoc method that refines fine-tuned models by selectively reverting parameter updates along principal directions of base weights. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Proposes a fine-tuning method rather than evaluation benchmarks or methodology.
Geometric Calibration of Supervised Fine-Tuning This paper presents a post-hoc method that refines fine-tuned models by selectively reverting parameter updates along principal directions of base weights. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Investigates fine-tuning weight geometry rather than autonomous agent systems.
Geometric Calibration of Supervised Fine-Tuning This paper presents a post-hoc method that refines fine-tuned models by selectively reverting parameter updates along principal directions of base weights. Preprint	Small and efficient models via arxiv · 15 Oct, 06:17	Not relevant Deals with weight subspace calibration rather than compression or efficient serving.
DTScribe: Generating Models for Digital Twin Reporting using Local Large Language Models The authors present an approach using local language models to automatically generate conceptual architectural models and reports for digital twins. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Focuses on digital twin documentation without addressing safety or reliability risks.
DTScribe: Generating Models for Digital Twin Reporting using Local Large Language Models The authors present an approach using local language models to automatically generate conceptual architectural models and reports for digital twins. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Uses LLM prompting loops rather than studying multi-step autonomous agent architectures.
DTScribe: Generating Models for Digital Twin Reporting using Local Large Language Models The authors present an approach using local language models to automatically generate conceptual architectural models and reports for digital twins. Preprint	LLMs for code via arxiv · 15 Oct, 06:17	Not relevant Generates conceptual architecture models for digital twins rather than software code.
DTScribe: Generating Models for Digital Twin Reporting using Local Large Language Models The authors present an approach using local language models to automatically generate conceptual architectural models and reports for digital twins. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Uses LLM-as-a-judge for an application rather than studying evaluation methodology.
To Align or Not To Align: Check Your COMPASS Before You Train This paper introduces a metric to estimate the compatibility of pretrained models before running costly alignment or checkpoint merging. Preprint	LLMs for code via arxiv · 15 Oct, 06:17	Not relevant Focuses on model alignment metrics rather than code generation or repair.
To Align or Not To Align: Check Your COMPASS Before You Train This paper introduces a metric to estimate the compatibility of pretrained models before running costly alignment or checkpoint merging. Preprint	Small and efficient models via arxiv · 15 Oct, 06:17	Not relevant Focuses on model compatibility diagnostics rather than distillation or quantization.
To Align or Not To Align: Check Your COMPASS Before You Train This paper introduces a metric to estimate the compatibility of pretrained models before running costly alignment or checkpoint merging. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Studies encoder alignment compatibility rather than agent tool use or planning.
To Align or Not To Align: Check Your COMPASS Before You Train This paper introduces a metric to estimate the compatibility of pretrained models before running costly alignment or checkpoint merging. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Evaluates cross-modal representation alignment rather than deployment safety risks.
To Align or Not To Align: Check Your COMPASS Before You Train This paper introduces a metric to estimate the compatibility of pretrained models before running costly alignment or checkpoint merging. Preprint	Retrieval-augmented generation (RAG) via arxiv · 15 Oct, 06:17	Not relevant Deals with pre-training encoder alignment, not external retrieval or RAG systems.
ST-GRU PredRNN-V2: A GRU-Based Reformulation of PredRNN-V2 for Video Prediction The authors reformulate a recurrent video prediction architecture using gated recurrent units to reduce parameters while maintaining spatiotemporal modeling. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Develops a video prediction GRU architecture, not an autonomous agent.
ST-GRU PredRNN-V2: A GRU-Based Reformulation of PredRNN-V2 for Video Prediction The authors reformulate a recurrent video prediction architecture using gated recurrent units to reduce parameters while maintaining spatiotemporal modeling. Preprint	Computer vision for industrial applications via arxiv · 15 Oct, 06:17	Not relevant Evaluates on Moving MNIST, not industrial defect detection or manufacturing inspection.
ST-GRU PredRNN-V2: A GRU-Based Reformulation of PredRNN-V2 for Video Prediction The authors reformulate a recurrent video prediction architecture using gated recurrent units to reduce parameters while maintaining spatiotemporal modeling. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Applies a model to existing benchmarks rather than building evaluation methods.
ST-GRU PredRNN-V2: A GRU-Based Reformulation of PredRNN-V2 for Video Prediction The authors reformulate a recurrent video prediction architecture using gated recurrent units to reduce parameters while maintaining spatiotemporal modeling. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Studies video prediction architectures rather than deployment safety or guardrails.
ST-GRU PredRNN-V2: A GRU-Based Reformulation of PredRNN-V2 for Video Prediction The authors reformulate a recurrent video prediction architecture using gated recurrent units to reduce parameters while maintaining spatiotemporal modeling. Preprint	Multimodal models via arxiv · 15 Oct, 06:17	Not relevant Focuses on spatiotemporal recurrent architectures for video frames without multimodal text integration.
Tiny Recursive Message Passing This paper applies recursive training and discrete deduction procedures to message-passing neural networks to solve constraint-reasoning tasks with few parameters. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Evaluates an architecture on puzzles rather than proposing evaluation benchmarks.
Tiny Recursive Message Passing This paper applies recursive training and discrete deduction procedures to message-passing neural networks to solve constraint-reasoning tasks with few parameters. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Studies algorithmic reasoning efficiency rather than safety or alignment guardrails.
Tiny Recursive Message Passing This paper applies recursive training and discrete deduction procedures to message-passing neural networks to solve constraint-reasoning tasks with few parameters. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Investigates message passing graph architectures, not LLM agent systems.
AMemOps: Evaluating Memory Updates in Long Audio Streams The authors introduce a benchmark to evaluate how audio models update, retain, or reject recurring acoustic sources in long streams. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Tests streaming audio memory operations rather than multi-step agent workflows.
AMemOps: Evaluating Memory Updates in Long Audio Streams The authors introduce a benchmark to evaluate how audio models update, retain, or reject recurring acoustic sources in long streams. Preprint	Small and efficient models via arxiv · 15 Oct, 06:17	Not relevant Evaluates memory updating mechanisms, not model compression or efficient serving.
AMemOps: Evaluating Memory Updates in Long Audio Streams The authors introduce a benchmark to evaluate how audio models update, retain, or reject recurring acoustic sources in long streams. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Focuses on memory update accuracy rather than safety risks or guardrails.
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. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Focuses on KV cache attention architecture, not multi-step agent planning or tools.
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. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Focuses on attention efficiency without addressing safety, jailbreaks, or reliability.
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. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Introduces a model architecture rather than evaluation benchmarks or methodology.
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. Preprint	Retrieval-augmented generation (RAG) via arxiv · 15 Oct, 06:17	Not relevant Proposes attention compression architecture rather than external document retrieval or grounding.

14,161–14,220 of 17,313← Previous Next →