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
Learning to GUIDE the Group: Self-Learning LLM Facilitation Across Multi-Party Tasks Presents a framework for developing language model facilitators that balance group dynamics, psychological safety, and task goals across multi-party conversations. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Focuses on multi-party facilitation rather than safety or guardrails.
AutoNovo: Agentic Program Search Discovers Transferable Nucleic Acid Design Algorithms Uses agentic program search to iteratively write and refine design algorithms for nucleic acid sequences that transfer across independent models. Preprint	Small and efficient models via arxiv · 15 Oct, 06:17	Not relevant Addresses biological sequence generation, not model compression.
AutoNovo: Agentic Program Search Discovers Transferable Nucleic Acid Design Algorithms Uses agentic program search to iteratively write and refine design algorithms for nucleic acid sequences that transfer across independent models. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Focuses on biological optimization rather than AI deployment safety.
AutoNovo: Agentic Program Search Discovers Transferable Nucleic Acid Design Algorithms Uses agentic program search to iteratively write and refine design algorithms for nucleic acid sequences that transfer across independent models. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Proposes a design method, not an LLM benchmark.
Reflected Proximal Dynamics for Accelerated High-Accuracy Log-Concave Sampling Proposes reflected proximal dynamics to combine exact target preservation with momentum-based acceleration in log-concave sampling. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Presents an optimization/sampling algorithm, not an evaluation benchmark.
Reflected Proximal Dynamics for Accelerated High-Accuracy Log-Concave Sampling Proposes reflected proximal dynamics to combine exact target preservation with momentum-based acceleration in log-concave sampling. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Theoretical sampling work unrelated to deployment safety.
Reflected Proximal Dynamics for Accelerated High-Accuracy Log-Concave Sampling Proposes reflected proximal dynamics to combine exact target preservation with momentum-based acceleration in log-concave sampling. Preprint	Small and efficient models via arxiv · 15 Oct, 06:17	Not relevant Deals with log-concave sampling theory, not model efficiency.
Reflected Proximal Dynamics for Accelerated High-Accuracy Log-Concave Sampling Proposes reflected proximal dynamics to combine exact target preservation with momentum-based acceleration in log-concave sampling. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Focuses on mathematical sampling theory, not AI agents.
Avoiding What Your Teacher Already Knows: Entity Familiarity for Deep-Search Distillation Measures teacher entity familiarity to filter out uninformative search trajectories when distilling deep-search capabilities into smaller agents. Preprint	LLMs for code via arxiv · 15 Oct, 06:17	Not relevant Not focused on coding assistants or software engineering.
Avoiding What Your Teacher Already Knows: Entity Familiarity for Deep-Search Distillation Measures teacher entity familiarity to filter out uninformative search trajectories when distilling deep-search capabilities into smaller agents. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Presents a distillation filtering method rather than an evaluation benchmark.
Avoiding What Your Teacher Already Knows: Entity Familiarity for Deep-Search Distillation Measures teacher entity familiarity to filter out uninformative search trajectories when distilling deep-search capabilities into smaller agents. Preprint	Multimodal models via arxiv · 15 Oct, 06:17	Not relevant Focuses on text entity graphs, not multimodal models.
Avoiding What Your Teacher Already Knows: Entity Familiarity for Deep-Search Distillation Measures teacher entity familiarity to filter out uninformative search trajectories when distilling deep-search capabilities into smaller agents. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Focuses on distillation data quality, not deployment safety.
ScreenASR-Bench: A Tiered Benchmark for Speech Recognition with Visual Context Introduces a tiered benchmark to assess how explicit and implicit visual evidence helps speech recognition models resolve ambiguous words on screens. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Focuses on ASR evaluation, not safety or guardrails.
ScreenASR-Bench: A Tiered Benchmark for Speech Recognition with Visual Context Introduces a tiered benchmark to assess how explicit and implicit visual evidence helps speech recognition models resolve ambiguous words on screens. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Benchmarks speech recognition, not autonomous agentic systems.
Hardware-Friendly Token-Group Activation Quantization for Low-Bit Mamba Super-Resolution Presents an activation quantization method that groups tokens via pre-scan features to improve low-bit post-training quantization for Mamba super-resolution models. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Addresses low-level quantization, not language model agents.
Hardware-Friendly Token-Group Activation Quantization for Low-Bit Mamba Super-Resolution Presents an activation quantization method that groups tokens via pre-scan features to improve low-bit post-training quantization for Mamba super-resolution models. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Focuses on quantization efficiency, not deployment safety or reliability.
Hardware-Friendly Token-Group Activation Quantization for Low-Bit Mamba Super-Resolution Presents an activation quantization method that groups tokens via pre-scan features to improve low-bit post-training quantization for Mamba super-resolution models. Preprint	Multimodal models via arxiv · 15 Oct, 06:17	Not relevant Focuses on vision super-resolution quantization, not multimodal models.
Hardware-Friendly Token-Group Activation Quantization for Low-Bit Mamba Super-Resolution Presents an activation quantization method that groups tokens via pre-scan features to improve low-bit post-training quantization for Mamba super-resolution models. Preprint	Computer vision for industrial applications via arxiv · 15 Oct, 06:17	Not relevant Addresses general super-resolution quantization, not industrial defect inspection.
Hardware-Friendly Token-Group Activation Quantization for Low-Bit Mamba Super-Resolution Presents an activation quantization method that groups tokens via pre-scan features to improve low-bit post-training quantization for Mamba super-resolution models. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Proposes a quantization method, not a benchmarking protocol.
A Survey of Flow Matching in Reinforcement Learning Surveys the use of flow-matching generative modeling approaches within reinforcement learning across policy parameterization and foundation model alignment. Peer-reviewed	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant A survey of RL algorithms, not an evaluation benchmark.
A Survey of Flow Matching in Reinforcement Learning Surveys the use of flow-matching generative modeling approaches within reinforcement learning across policy parameterization and foundation model alignment. Peer-reviewed	Small and efficient models via arxiv · 15 Oct, 06:17	Not relevant Addresses reinforcement learning policies rather than small language models.
A Survey of Flow Matching in Reinforcement Learning Surveys the use of flow-matching generative modeling approaches within reinforcement learning across policy parameterization and foundation model alignment. Peer-reviewed	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Surveys flow matching in RL, not deployment safety or guardrails.
A Survey of Flow Matching in Reinforcement Learning Surveys the use of flow-matching generative modeling approaches within reinforcement learning across policy parameterization and foundation model alignment. Peer-reviewed	AI agents via arxiv · 15 Oct, 06:17	Not relevant Surveys generative modeling in RL, not multi-step LLM agents.
agent-risk: Risk Assessment for Changes to AI Agents Provides a framework for assessing risks of proposed changes to deployed AI agents by selectively replaying execution logs and saved states. Preprint	LLMs for code via arxiv · 15 Oct, 06:17	Not relevant Focuses on agent policy replay, not software development code.
agent-risk: Risk Assessment for Changes to AI Agents Provides a framework for assessing risks of proposed changes to deployed AI agents by selectively replaying execution logs and saved states. Preprint	Multimodal models via arxiv · 15 Oct, 06:17	Not relevant Does not deal with multimodal inputs or vision-language models.
agent-risk: Risk Assessment for Changes to AI Agents Provides a framework for assessing risks of proposed changes to deployed AI agents by selectively replaying execution logs and saved states. Preprint	Retrieval-augmented generation (RAG) via arxiv · 15 Oct, 06:17	Not relevant Does not address retrieval-augmented generation.
agent-risk: Risk Assessment for Changes to AI Agents Provides a framework for assessing risks of proposed changes to deployed AI agents by selectively replaying execution logs and saved states. Preprint	Small and efficient models via arxiv · 15 Oct, 06:17	Not relevant Does not address model compression or efficient inference.
REVEAL: Reverse Engineering Agentic Latents from Frontier-Model Traces Extracts reusable local control signals from frontier model traces to guide smaller language model agents without extra target-model training. Preprint	Retrieval-augmented generation (RAG) via arxiv · 15 Oct, 06:17	Not relevant Does not cover retrieval-augmented generation from external knowledge stores.
REVEAL: Reverse Engineering Agentic Latents from Frontier-Model Traces Extracts reusable local control signals from frontier model traces to guide smaller language model agents without extra target-model training. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Focuses on agent steering and capability transfer, not safety.
REVEAL: Reverse Engineering Agentic Latents from Frontier-Model Traces Extracts reusable local control signals from frontier model traces to guide smaller language model agents without extra target-model training. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Proposes an agent steering method, not an evaluation benchmark.
REVEAL: Reverse Engineering Agentic Latents from Frontier-Model Traces Extracts reusable local control signals from frontier model traces to guide smaller language model agents without extra target-model training. Preprint	Multimodal models via arxiv · 15 Oct, 06:17	Not relevant Focuses on textual agent trajectories, not multimodal vision-language models.
REVEAL: Reverse Engineering Agentic Latents from Frontier-Model Traces Extracts reusable local control signals from frontier model traces to guide smaller language model agents without extra target-model training. Preprint	LLMs for code via arxiv · 15 Oct, 06:17	Not relevant Not focused on code generation or software engineering tasks.
SLDBench v2: Active Scaling Law Discovery Expands a benchmark to evaluate whether language model agents can actively select experiments to discover reliable model scaling laws under compute budgets. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Focuses on discovery benchmarks, not safety or guardrails.
SLDBench v2: Active Scaling Law Discovery Expands a benchmark to evaluate whether language model agents can actively select experiments to discover reliable model scaling laws under compute budgets. Preprint	Small and efficient models via arxiv · 15 Oct, 06:17	Not relevant Studies scaling law discovery rather than efficient inference or distillation.
Memory-Efficient Integer-Only Training of Deep Neural Networks Introduces an INT8 integer-only training algorithm using composable building blocks that removes floating-point arithmetic across all training operations. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Focuses on low-precision training mechanics, not evaluation benchmarks.
Memory-Efficient Integer-Only Training of Deep Neural Networks Introduces an INT8 integer-only training algorithm using composable building blocks that removes floating-point arithmetic across all training operations. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Does not study safety, alignment, or deployment reliability.
Memory-Efficient Integer-Only Training of Deep Neural Networks Introduces an INT8 integer-only training algorithm using composable building blocks that removes floating-point arithmetic across all training operations. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Addresses integer-only training, not AI agents.
FON: A Draft-Revise Dual-Organ Language Model without Sequence-Level Quadratic Attention Proposes an attention-free language model architecture separating drafting from multi-scale prefix revision to bypass sequence-level quadratic attention costs. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Proposes a model architecture, not benchmark methodology.
FON: A Draft-Revise Dual-Organ Language Model without Sequence-Level Quadratic Attention Proposes an attention-free language model architecture separating drafting from multi-scale prefix revision to bypass sequence-level quadratic attention costs. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Focuses on neural architecture, not agent workflows.
FON: A Draft-Revise Dual-Organ Language Model without Sequence-Level Quadratic Attention Proposes an attention-free language model architecture separating drafting from multi-scale prefix revision to bypass sequence-level quadratic attention costs. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Does not address safety, guardrails, or hallucination.
Narrative Paradoxes in LLM Inference: Shaping Reasoning Trajectories toward Misalignment Develops a reasoning-based attack that pairs malicious actors with benign contexts to guide language models along harmful multi-step reasoning trajectories. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Studies jailbreak trajectories, not autonomous tool-use or multi-agent systems.
Narrative Paradoxes in LLM Inference: Shaping Reasoning Trajectories toward Misalignment Develops a reasoning-based attack that pairs malicious actors with benign contexts to guide language models along harmful multi-step reasoning trajectories. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Introduces an attack method rather than standard evaluation frameworks.
ReTAC: Residual Test-Time Agentic Computation for Video Spatial Reasoning Presents a multi-agent framework that allocates test-time computation only to unresolved steps for video spatial reasoning tasks. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Focuses on spatial task execution, not safety or guardrails.
ReTAC: Residual Test-Time Agentic Computation for Video Spatial Reasoning Presents a multi-agent framework that allocates test-time computation only to unresolved steps for video spatial reasoning tasks. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Proposes an agent framework, not a benchmark or evaluation methodology.
HyWe: Unifying Search and Operation for Real-World Hybrid Web Tasks Introduces a 35B model, trajectory pipeline, and benchmark unifying search and browser interaction within a single web agent. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Focuses on web agent capability rather than alignment or guardrails.
BehaviorFormer: Discovering Behavioral Patterns via Clusterable Next-Token Prediction Combines next-token prediction transformers with deep clustering to discover interpretable behavioral patterns in open-ended game trajectories. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Analyzes game player trajectories rather than AI deployment safety risks.
BehaviorFormer: Discovering Behavioral Patterns via Clusterable Next-Token Prediction Combines next-token prediction transformers with deep clustering to discover interpretable behavioral patterns in open-ended game trajectories. Preprint	Small and efficient models via arxiv · 15 Oct, 06:17	Not relevant Focuses on behavioral representation learning rather than model compression.
BehaviorFormer: Discovering Behavioral Patterns via Clusterable Next-Token Prediction Combines next-token prediction transformers with deep clustering to discover interpretable behavioral patterns in open-ended game trajectories. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Presents a behavioral clustering technique rather than an LLM evaluation benchmark.
BehaviorFormer: Discovering Behavioral Patterns via Clusterable Next-Token Prediction Combines next-token prediction transformers with deep clustering to discover interpretable behavioral patterns in open-ended game trajectories. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Discovers human player patterns in games, not autonomous agentic systems.
When a Conceptor Complement Becomes Gain: Operator Geometry and Behavioral Attribution in Language Models Analyzes activation steering interventions in language models by isolating the effects of covariance filtering from target gain and displacement. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Studies steering matrix geometry rather than broad LLM evaluation benchmarks.
When a Conceptor Complement Becomes Gain: Operator Geometry and Behavioral Attribution in Language Models Analyzes activation steering interventions in language models by isolating the effects of covariance filtering from target gain and displacement. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Analyzes model steering mechanisms, not multi-step autonomous planning agents.
Explicitly Aligned Multi-View Breast Cancer Risk Prediction from Variable-Length Screening Histories Presents a breast cancer risk prediction framework that explicitly aligns longitudinal examinations across multiple mammography views. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Studies medical mammography analysis, not autonomous planning agents.
Explicitly Aligned Multi-View Breast Cancer Risk Prediction from Variable-Length Screening Histories Presents a breast cancer risk prediction framework that explicitly aligns longitudinal examinations across multiple mammography views. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Addresses medical risk prediction rather than LLM benchmarking standards.
Explicitly Aligned Multi-View Breast Cancer Risk Prediction from Variable-Length Screening Histories Presents a breast cancer risk prediction framework that explicitly aligns longitudinal examinations across multiple mammography views. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Addresses medical diagnostic prediction rather than AI deployment guardrails.
Explicitly Aligned Multi-View Breast Cancer Risk Prediction from Variable-Length Screening Histories Presents a breast cancer risk prediction framework that explicitly aligns longitudinal examinations across multiple mammography views. Preprint	Multimodal models via arxiv · 15 Oct, 06:17	Not relevant Concerns medical image alignment rather than general multimodal language models.
Explicitly Aligned Multi-View Breast Cancer Risk Prediction from Variable-Length Screening Histories Presents a breast cancer risk prediction framework that explicitly aligns longitudinal examinations across multiple mammography views. Preprint	Computer vision for industrial applications via arxiv · 15 Oct, 06:17	Not relevant Focuses on medical screening, not industrial manufacturing defect detection.
BAT3R: Robust Online 3D Reconstruction with Bayesian Adaptive State Updates Introduces a training-free Bayesian state-update framework to reduce error accumulation and representational drift in streaming 3D reconstruction models. Preprint	Computer vision for industrial applications via arxiv · 15 Oct, 06:17	Not relevant Deals with general 3D video reconstruction, not industrial visual inspection.
BAT3R: Robust Online 3D Reconstruction with Bayesian Adaptive State Updates Introduces a training-free Bayesian state-update framework to reduce error accumulation and representational drift in streaming 3D reconstruction models. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:17	Not relevant Focuses on reconstruction drift rather than deployment safety guardrails.
BAT3R: Robust Online 3D Reconstruction with Bayesian Adaptive State Updates Introduces a training-free Bayesian state-update framework to reduce error accumulation and representational drift in streaming 3D reconstruction models. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:17	Not relevant Introduces a 3D reconstruction framework, not an LLM benchmark.
BAT3R: Robust Online 3D Reconstruction with Bayesian Adaptive State Updates Introduces a training-free Bayesian state-update framework to reduce error accumulation and representational drift in streaming 3D reconstruction models. Preprint	AI agents via arxiv · 15 Oct, 06:17	Not relevant Focuses on 3D reconstruction algorithms, not autonomous LLM agents.

12,781–12,840 of 17,313← Previous Next →