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
TabMF: Learning Dataset Representations with Tabular Foundation Models Repurposes pretrained tabular foundation models into supervised dataset encoders using dataset-level readouts and contrastive post-training. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Does not address AI safety, jailbreaks, hallucination mitigation, or deployment guardrails.
TabMF: Learning Dataset Representations with Tabular Foundation Models Repurposes pretrained tabular foundation models into supervised dataset encoders using dataset-level readouts and contrastive post-training. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Focuses on tabular foundation model embeddings, not model compression or efficient inference techniques.
TabMF: Learning Dataset Representations with Tabular Foundation Models Repurposes pretrained tabular foundation models into supervised dataset encoders using dataset-level readouts and contrastive post-training. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Studies dataset representation learning rather than LLM benchmarks or evaluation methods.
Shared Canonical Component Removal Beyond Expert Views Presents Shared Canonical Component Removal to quantify predictive information retained by raw machine learning representations beyond predefined expert features. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Investigates feature representation auditing in clinical models, not LLM evaluation or benchmarks.
Shared Canonical Component Removal Beyond Expert Views Presents Shared Canonical Component Removal to quantify predictive information retained by raw machine learning representations beyond predefined expert features. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Presents an audit framework for expert feature extraction rather than deployed AI safety guardrails.
Shared Canonical Component Removal Beyond Expert Views Presents Shared Canonical Component Removal to quantify predictive information retained by raw machine learning representations beyond predefined expert features. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Does not address model compression, small language models, or efficient serving.
On the Design Space of Continuous Diffusion Language Models Investigates training mechanisms in continuous diffusion language models, proposing design modifications for self-conditioning and embedding geometry. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Analyzes continuous diffusion architecture design rather than benchmarking or evaluating language models.
On the Design Space of Continuous Diffusion Language Models Investigates training mechanisms in continuous diffusion language models, proposing design modifications for self-conditioning and embedding geometry. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Focuses on diffusion model architecture and sampling efficiency, not safety, guardrails, or alignment.
Specialists Learn Faster: Knowledge Organization in Pre-training Shapes Knowledge Update Efficiency Examines how chronological versus shuffled pre-training knowledge sequencing influences the sample efficiency of subsequent model knowledge updates. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Investigates pre-training knowledge update dynamics rather than safety, red-teaming, or deployment reliability.
Specialists Learn Faster: Knowledge Organization in Pre-training Shapes Knowledge Update Efficiency Examines how chronological versus shuffled pre-training knowledge sequencing influences the sample efficiency of subsequent model knowledge updates. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Focuses on pre-training data ordering dynamics, not LLM benchmark design or evaluation frameworks.
Returning to Development States: When Full Sandbox Snapshots Help Coding Agents Evaluates full sandbox snapshot exploration for coding agents, finding benefits primarily when recovery requires restoring runtime dependencies outside Git repositories. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant It does not introduce or study general LLM evaluation benchmarks or assessment methodology.
Returning to Development States: When Full Sandbox Snapshots Help Coding Agents Evaluates full sandbox snapshot exploration for coding agents, finding benefits primarily when recovery requires restoring runtime dependencies outside Git repositories. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Deals with task retry efficiency for coding agents rather than deployment safety, alignment, or guardrails.
Returning to Development States: When Full Sandbox Snapshots Help Coding Agents Evaluates full sandbox snapshot exploration for coding agents, finding benefits primarily when recovery requires restoring runtime dependencies outside Git repositories. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Does not study model compression, distillation, quantization, or small on-device models.
Inference-Time Steering Across Scaffolds and Epitopes for Sample-Efficient De Novo Nanobody Design Applies dynamic budget allocation algorithms at inference time to select scaffolds and epitopes during de novo nanobody generative design. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Computational biology nanobody generation paper, not AI safety or deployment reliability.
Inference-Time Steering Across Scaffolds and Epitopes for Sample-Efficient De Novo Nanobody Design Applies dynamic budget allocation algorithms at inference time to select scaffolds and epitopes during de novo nanobody generative design. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Covers molecular design inference steering, not LLM evaluation benchmarks.
Inference-Time Steering Across Scaffolds and Epitopes for Sample-Efficient De Novo Nanobody Design Applies dynamic budget allocation algorithms at inference time to select scaffolds and epitopes during de novo nanobody generative design. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Focuses on compute steering for molecular generation, not small or efficient language models.
Finite-Sample Guarantees for Fully Discrete Conditional Diffusion Models Derives finite-sample convergence guarantees for fully discrete conditional diffusion models with finitely many denoising steps under smooth density assumptions. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Mathematical finite-sample theory for conditional diffusion, not practical small model efficiency.
Finite-Sample Guarantees for Fully Discrete Conditional Diffusion Models Derives finite-sample convergence guarantees for fully discrete conditional diffusion models with finitely many denoising steps under smooth density assumptions. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Theoretical statistical estimation bounds for diffusion models, not LLM evaluation.
Finite-Sample Guarantees for Fully Discrete Conditional Diffusion Models Derives finite-sample convergence guarantees for fully discrete conditional diffusion models with finitely many denoising steps under smooth density assumptions. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Statistical theoretical guarantees, not operational AI safety or deployment reliability.
Partially Informed Stochastic Optimization via Zeroth-Order Residual Tracking Introduces PISO, an optimization framework that combines partially available gradients with zeroth-order residual tracking to reduce variance from random probing. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant General stochastic optimization algorithms, not AI safety or deployment robustness.
Partially Informed Stochastic Optimization via Zeroth-Order Residual Tracking Introduces PISO, an optimization framework that combines partially available gradients with zeroth-order residual tracking to reduce variance from random probing. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Presents theoretical stochastic optimization algorithms, not LLM benchmarking.
Partially Informed Stochastic Optimization via Zeroth-Order Residual Tracking Introduces PISO, an optimization framework that combines partially available gradients with zeroth-order residual tracking to reduce variance from random probing. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Optimization theory paper, not focused on efficient language model architectures or compression.
Prediction Is Not a Sufficient Memory State Analyzes the dimensional requirements of exact recurrent memory states for cumulative learning under convex residual loss functions. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Focuses on theoretical memory state dimension in online learning, not efficient model deployment.
Prediction Is Not a Sufficient Memory State Analyzes the dimensional requirements of exact recurrent memory states for cumulative learning under convex residual loss functions. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Theoretical learning theory, not safety, guardrails, or red-teaming.
Prediction Is Not a Sufficient Memory State Analyzes the dimensional requirements of exact recurrent memory states for cumulative learning under convex residual loss functions. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Theoretical characterization of cumulative learning memory states, not LLM evaluation.
Hitting-Time Bounds for Asynchronous Q-Learning Establishes finite-time error bounds for single-trajectory asynchronous Q-learning that incorporate the hitting time of the underlying Markov decision process. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Pure theoretical analysis of Q-learning, not operational AI safety or reliability.
Hitting-Time Bounds for Asynchronous Q-Learning Establishes finite-time error bounds for single-trajectory asynchronous Q-learning that incorporate the hitting time of the underlying Markov decision process. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Theoretical convergence bounds for asynchronous Q-learning, not LLM benchmarks.
Hitting-Time Bounds for Asynchronous Q-Learning Establishes finite-time error bounds for single-trajectory asynchronous Q-learning that incorporate the hitting time of the underlying Markov decision process. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Reinforcement learning theory, not model compression or efficient serving.
Designing Protein Structures from Natural Language Introduces Prompteina, an all-atom generative model and dataset for generating de novo protein structures conditioned on natural language prompts. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Focuses on protein structure generation from text prompts, not LLM evaluation methods.
Designing Protein Structures from Natural Language Introduces Prompteina, an all-atom generative model and dataset for generating de novo protein structures conditioned on natural language prompts. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Applies text conditioning to protein design models, not small language model efficiency.
Designing Protein Structures from Natural Language Introduces Prompteina, an all-atom generative model and dataset for generating de novo protein structures conditioned on natural language prompts. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Addresses protein structure design, not AI safety, guardrails, or alignment.
DyssomniaNN: Multi-Scale Spiking Neural Modeling of Human Sleep Disorders and Perceived Restedness Develops a multi-scale biophysical spiking neural network model to simulate sleep-wake regulation and discrepancies between objective sleep and subjective restedness. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Deals with computational neuroscience and sleep modeling, not LLM evaluation.
DyssomniaNN: Multi-Scale Spiking Neural Modeling of Human Sleep Disorders and Perceived Restedness Develops a multi-scale biophysical spiking neural network model to simulate sleep-wake regulation and discrepancies between objective sleep and subjective restedness. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Computational neuroscience research, not AI deployment safety or jailbreak defense.
DyssomniaNN: Multi-Scale Spiking Neural Modeling of Human Sleep Disorders and Perceived Restedness Develops a multi-scale biophysical spiking neural network model to simulate sleep-wake regulation and discrepancies between objective sleep and subjective restedness. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Biophysical spiking neural network model for sleep disorders, not efficient LLM serving.
Goal-Oriented Bundle Management and Flow Control for Deep-Space Communications via Reinforcement Learning Applies reinforcement learning to manage bundle scheduling and flow control at relay nodes in deep-space communication networks. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Focuses on deep-space communication networking via RL, not LLM evaluation.
Goal-Oriented Bundle Management and Flow Control for Deep-Space Communications via Reinforcement Learning Applies reinforcement learning to manage bundle scheduling and flow control at relay nodes in deep-space communication networks. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Addresses network bundle protocol scheduling, not small or efficient language models.
Goal-Oriented Bundle Management and Flow Control for Deep-Space Communications via Reinforcement Learning Applies reinforcement learning to manage bundle scheduling and flow control at relay nodes in deep-space communication networks. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Addresses space communication protocols, not AI safety or deployment guardrails.
KL Before and After Mixing: A Round-Complexity Separation in Parallel Sampling Analyzes round complexity in parallel coordinate samplers, proving a separation between schedule-averaged and output-law Kullback-Leibler divergences. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Theoretical probability bounds on sampling, not deployment safety or reliability.
KL Before and After Mixing: A Round-Complexity Separation in Parallel Sampling Analyzes round complexity in parallel coordinate samplers, proving a separation between schedule-averaged and output-law Kullback-Leibler divergences. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Theoretical analysis of parallel sampling divergences, not LLM benchmarks or evaluation.
KL Before and After Mixing: A Round-Complexity Separation in Parallel Sampling Analyzes round complexity in parallel coordinate samplers, proving a separation between schedule-averaged and output-law Kullback-Leibler divergences. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Focuses on theoretical sampling complexity, not small language models or efficient deployment.
Constructive Bounds for the Topological Expressivity of Neural Networks Proves upper bounds on the topological expressivity of neural networks measured by the Betti numbers of their output spaces. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Theoretical topology paper, not AI safety, robustness, or deployment guardrails.
Constructive Bounds for the Topological Expressivity of Neural Networks Proves upper bounds on the topological expressivity of neural networks measured by the Betti numbers of their output spaces. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Pure theory of ReLU network expressivity; not about small model efficiency or distillation.
Constructive Bounds for the Topological Expressivity of Neural Networks Proves upper bounds on the topological expressivity of neural networks measured by the Betti numbers of their output spaces. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Theoretical math paper on topological expressivity, not LLM evaluation or benchmarks.
Towards Unified Evaluation of Prompt Enhancers for Video Generation Introduces PEBench, a benchmark and evidence-grounded framework to directly evaluate prompt enhancers for video generation without rendering full videos. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Evaluates prompt enhancers for video generation, not industrial vision applications.
Streaming-Aware Diffusion for Real-Time Video Super-Resolution via Cross-Step Attention The paper proposes a streaming video super-resolution framework reusing intermediate latent diffusion features across frames via cross-step attention. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Focuses on real-time video super-resolution, not industrial inspection or defect detection.
Streaming-Aware Diffusion for Real-Time Video Super-Resolution via Cross-Step Attention The paper proposes a streaming video super-resolution framework reusing intermediate latent diffusion features across frames via cross-step attention. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Addresses video super-resolution via diffusion models, not multimodal models integrating text or other modalities.
Dino Forcing Flow Models: Do not denoise what you can predict The paper proposes predicting pretrained DINO representations directly to guide flow matching models, avoiding a secondary ODE denoising process. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Focuses on general generative image flow models, not industrial visual inspection.
Dino Forcing Flow Models: Do not denoise what you can predict The paper proposes predicting pretrained DINO representations directly to guide flow matching models, avoiding a secondary ODE denoising process. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Covers unimodal image generation and flow matching rather than multimodal cross-modality models.
Memory Forcing: Attendable Mid-Horizon History for Streaming Video Generation The paper introduces Memory Forcing, partitioning cache into archive and working banks to prevent mid-horizon history loss in streaming video generation. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Proposes streaming video diffusion architectures, not industrial vision or inspection.
From Video Clips to Creation Trajectory: Sora100K for AI-Native Video Creation The paper introduces Sora100K, a dataset structuring AI-native video creation workflows into multi-turn creation trajectories and editing operations. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Applies to generative video creation workflows, not industrial computer vision tasks.
Skill-V: Verifiable Self-Evolving Skill Library for Interactive Agents The paper introduces Skill-V, a verifiable agent skill library that treats skills as falsifiable contracts tested against environment execution outcomes. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Focuses on interactive agent skill libraries, not industrial vision or defect detection.
Skill-V: Verifiable Self-Evolving Skill Library for Interactive Agents The paper introduces Skill-V, a verifiable agent skill library that treats skills as falsifiable contracts tested against environment execution outcomes. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Does not address multimodal reasoning across text, images, video, or audio.
Phase-aware video generation for physics-grounded dynamics and interactions The paper proposes PAVG, a dual-branch video generator modeling distinct solid-gas dynamics and cross-attention interactions for physically grounded motion. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Focuses on generative video modeling of solid-gas dynamics, not multimodal reasoning.
Phase-aware video generation for physics-grounded dynamics and interactions The paper proposes PAVG, a dual-branch video generator modeling distinct solid-gas dynamics and cross-attention interactions for physically grounded motion. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Addresses physics-grounded video generation, not industrial visual inspection or robotics.
Seek-and-View Reasoning for Multi-View Spatial Understanding The paper proposes Vantage, a training-free framework pairing VLMs with 3D models to plan and synthesize relevant viewpoints for spatial reasoning. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Focuses on general multi-view spatial reasoning benchmarks, not industrial vision.
Fast Pose Tracking of Rigid Objects with Compact Pose Graph Optimization The paper proposes a lightweight pose-graph tracking module using relative pose constraints and geometric uncertainty for real-time 6D object tracking. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Focuses on geometric pose tracking and optimization, not multimodal models.
Open-Vocabulary Audio-Visual Event Localization via Complex-Valued Fusion The paper proposes a complex-valued neural network to fuse audio and visual similarity streams for open-vocabulary audio-visual event localization. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Does not address deployment safety, robustness attacks, or guardrails.
Open-Vocabulary Audio-Visual Event Localization via Complex-Valued Fusion The paper proposes a complex-valued neural network to fuse audio and visual similarity streams for open-vocabulary audio-visual event localization. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Focuses on audio-visual event localization benchmarks, not industrial operations.
Open-Vocabulary Audio-Visual Event Localization via Complex-Valued Fusion The paper proposes a complex-valued neural network to fuse audio and visual similarity streams for open-vocabulary audio-visual event localization. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Proposes a new fusion architecture, not an evaluation framework or benchmark study.
Open-Vocabulary Audio-Visual Event Localization via Complex-Valued Fusion The paper proposes a complex-valued neural network to fuse audio and visual similarity streams for open-vocabulary audio-visual event localization. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not focus on model compression, quantization, or inference acceleration.

15,301–15,360 of 17,313← Previous Next →