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
Occupancy Modeling via Drifting: One-Step Generation Enables Direct Policy Extraction Proposes a one-step generative approach to model occupancy for simplified policy extraction in reinforcement learning. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 15:06	Not relevant Addresses reinforcement learning policy extraction, not LLM benchmarking.
Occupancy Modeling via Drifting: One-Step Generation Enables Direct Policy Extraction Proposes a one-step generative approach to model occupancy for simplified policy extraction in reinforcement learning. Preprint	Small and efficient models via arxiv · 15 Oct, 15:06	Not relevant Focuses on RL occupancy models rather than small language models.
Few-Shot Adaptation for Anomalous Sound Detection at Low False-Positive Rate Improves industrial sound anomaly detection by adapting unsupervised models with few fault samples to reduce false alarms. Peer-reviewed	AI safety and reliability in deployment via arxiv · 15 Oct, 15:06	Not relevant Focuses on industrial audio anomaly detection rather than AI system safety.
Few-Shot Adaptation for Anomalous Sound Detection at Low False-Positive Rate Improves industrial sound anomaly detection by adapting unsupervised models with few fault samples to reduce false alarms. Peer-reviewed	Small and efficient models via arxiv · 15 Oct, 15:06	Not relevant Addresses acoustic anomaly detection algorithms, not small language model efficiency.
Few-Shot Adaptation for Anomalous Sound Detection at Low False-Positive Rate Improves industrial sound anomaly detection by adapting unsupervised models with few fault samples to reduce false alarms. Peer-reviewed	LLM evaluation and benchmarks via arxiv · 15 Oct, 15:06	Not relevant Studies acoustic machinery fault detection, not language model evaluation.
When Non-Adversarial Noise Hurts Learning: Benchmarking Skill Self-Evolution in Agents Introduces a benchmark to evaluate how language model agents learn and self-evolve reusable skills when exposed to non-adversarial noise. Preprint	Small and efficient models via arxiv · 15 Oct, 15:06	Not relevant Focuses on agent skill evolution and noise robustness, not model compression.
Beyond DFT electronic band structures via dual-space orbital learning Predicts high-level electronic band structures of materials using orbital information combined with low-cost baseline calculations. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 15:06	Not relevant Application to material science and quantum mechanics, not LLM evaluation.
Beyond DFT electronic band structures via dual-space orbital learning Predicts high-level electronic band structures of materials using orbital information combined with low-cost baseline calculations. Preprint	Small and efficient models via arxiv · 15 Oct, 15:06	Not relevant Deals with electronic band structure prediction, not small language models.
Beyond DFT electronic band structures via dual-space orbital learning Predicts high-level electronic band structures of materials using orbital information combined with low-cost baseline calculations. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 15:06	Not relevant Physics-informed material property prediction, not AI safety or jailbreak prevention.
Convergence Rates of Gaussian Process Thompson Sampling for Efficient Global Optimization Establishes theoretical convergence bounds for Gaussian process Thompson sampling under noiseless conditions in global optimization. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 15:06	Not relevant Addresses theoretical black-box optimization, not deployment safety or monitoring.
Convergence Rates of Gaussian Process Thompson Sampling for Efficient Global Optimization Establishes theoretical convergence bounds for Gaussian process Thompson sampling under noiseless conditions in global optimization. Preprint	Small and efficient models via arxiv · 15 Oct, 15:06	Not relevant Focuses on Gaussian process sampling bounds, not efficient LLM serving.
Convergence Rates of Gaussian Process Thompson Sampling for Efficient Global Optimization Establishes theoretical convergence bounds for Gaussian process Thompson sampling under noiseless conditions in global optimization. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 15:06	Not relevant Theoretical study of Bayesian optimization convergence rates, not LLM evaluation.
MorphingNet: Symbolic Regression Where the Network Is the Formula Treats symbolic regression as a continuous optimization problem using neural units that morph dynamically into mathematical formulas. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 15:06	Not relevant Presents a neural architecture for symbolic regression, not LLM benchmarks.
MorphingNet: Symbolic Regression Where the Network Is the Formula Treats symbolic regression as a continuous optimization problem using neural units that morph dynamically into mathematical formulas. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 15:06	Not relevant Focuses on scientific formula recovery, not LLM safety or deployment risks.
MorphingNet: Symbolic Regression Where the Network Is the Formula Treats symbolic regression as a continuous optimization problem using neural units that morph dynamically into mathematical formulas. Preprint	Small and efficient models via arxiv · 15 Oct, 15:06	Not relevant Studies continuous mathematical formula discovery rather than small language models.
LatentLoop: Scaling Looped MoE Models with Asymmetric Compute Allocation Allocates computation unevenly across passes in looped mixture-of-experts models to improve efficiency without losing performance. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 15:06	Not relevant Proposes an architectural design, not a language model evaluation method.
LatentLoop: Scaling Looped MoE Models with Asymmetric Compute Allocation Allocates computation unevenly across passes in looped mixture-of-experts models to improve efficiency without losing performance. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 15:06	Not relevant Focuses on compute allocation efficiency rather than safety or reliability.
Minimax estimation of the pseudo-spectral gap of continuous Markov chains, with applications to empirical concentration bounds The paper develops a minimax-optimal estimator for the pseudo-spectral gap of continuous-state Markov chains from a single observed trajectory. Preprint	Small and efficient models via arxiv · 15 Oct, 15:06	Not relevant Examines theoretical concentration bounds rather than efficient model inference.
Minimax estimation of the pseudo-spectral gap of continuous Markov chains, with applications to empirical concentration bounds The paper develops a minimax-optimal estimator for the pseudo-spectral gap of continuous-state Markov chains from a single observed trajectory. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 15:06	Not relevant Focuses on statistical Markov chain theory, not language model evaluation.
Minimax estimation of the pseudo-spectral gap of continuous Markov chains, with applications to empirical concentration bounds The paper develops a minimax-optimal estimator for the pseudo-spectral gap of continuous-state Markov chains from a single observed trajectory. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 15:06	Not relevant Theoretical statistics paper with no focus on AI deployment safety.
Outperforming 1000 Euclidean Layers With Hyperbolic Goal-Conditioned RL The paper demonstrates that hyperbolic contrastive reinforcement learning achieves strong navigation performance with much shallower networks than Euclidean models. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 15:06	Not relevant Focuses on RL navigation dynamics, not AI safety or deployment robustness.
Outperforming 1000 Euclidean Layers With Hyperbolic Goal-Conditioned RL The paper demonstrates that hyperbolic contrastive reinforcement learning achieves strong navigation performance with much shallower networks than Euclidean models. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 15:06	Not relevant Investigates hyperbolic reinforcement learning architectures, not LLM evaluation benchmarks.
Outperforming 1000 Euclidean Layers With Hyperbolic Goal-Conditioned RL The paper demonstrates that hyperbolic contrastive reinforcement learning achieves strong navigation performance with much shallower networks than Euclidean models. Preprint	Small and efficient models via arxiv · 15 Oct, 15:06	Not relevant Applies hyperbolic geometry to reinforcement learning navigation, not LLM efficiency.
ChemArgus: Automated Scientific Diagnostics for Lab-Grade Multimodal Chemistry Reasoning: A Frontier Benchmark The paper presents a multimodal benchmark for competition-level chemistry that grades reasoning step-by-step with partial credit. Preprint	Small and efficient models via arxiv · 15 Oct, 15:06	Not relevant Focuses on scientific reasoning evaluation, not model compression or efficiency.
ChemArgus: Automated Scientific Diagnostics for Lab-Grade Multimodal Chemistry Reasoning: A Frontier Benchmark The paper presents a multimodal benchmark for competition-level chemistry that grades reasoning step-by-step with partial credit. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 15:06	Not relevant Proposes an academic benchmark rather than deployment safety or guardrail frameworks.
Weighted Adjoint Matching: Finetuning Flow Matching Models with Time-Dependent Control Cost The paper introduces a time-dependent control-cost weighting method to fine-tune diffusion and flow-matching models with reward guidance. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 15:06	Not relevant Studies continuous generative path regularization, not deployment safety or jailbreaks.
Weighted Adjoint Matching: Finetuning Flow Matching Models with Time-Dependent Control Cost The paper introduces a time-dependent control-cost weighting method to fine-tune diffusion and flow-matching models with reward guidance. Preprint	Small and efficient models via arxiv · 15 Oct, 15:06	Not relevant Focuses on diffusion model fine-tuning control, not efficient model architectures.
Weighted Adjoint Matching: Finetuning Flow Matching Models with Time-Dependent Control Cost The paper introduces a time-dependent control-cost weighting method to fine-tune diffusion and flow-matching models with reward guidance. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 15:06	Not relevant Presents a fine-tuning method for diffusion models, not LLM benchmarks.
Sequential Learning-Based Physics-Informed Neural Network Modeling for Parameter Estimation on a Power Converter The paper proposes a staged physics-informed neural network to estimate electrical parameters of a direct-current buck power converter. Preprint	Small and efficient models via arxiv · 15 Oct, 15:06	Not relevant Focuses on power converter modeling rather than efficient LLM inference.
Sequential Learning-Based Physics-Informed Neural Network Modeling for Parameter Estimation on a Power Converter The paper proposes a staged physics-informed neural network to estimate electrical parameters of a direct-current buck power converter. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 15:06	Not relevant Applies PINNs to electrical engineering, unrelated to AI deployment safety.
Sequential Learning-Based Physics-Informed Neural Network Modeling for Parameter Estimation on a Power Converter The paper proposes a staged physics-informed neural network to estimate electrical parameters of a direct-current buck power converter. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 15:06	Not relevant Examines physics-informed neural networks for circuits, not LLM evaluation.
GBTSVR: Twin Support Vector Regression via Granular Ball Computing The paper develops twin support vector regression methods using granular ball computing to improve robustness against noise and outliers. Preprint	Small and efficient models via arxiv · 15 Oct, 15:06	Not relevant Proposes a support vector regression variant, not small or efficient LLMs.
GBTSVR: Twin Support Vector Regression via Granular Ball Computing The paper develops twin support vector regression methods using granular ball computing to improve robustness against noise and outliers. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 15:06	Not relevant Studies traditional tabular regression robustness, not LLM safety or guardrails.
GBTSVR: Twin Support Vector Regression via Granular Ball Computing The paper develops twin support vector regression methods using granular ball computing to improve robustness against noise and outliers. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 15:06	Not relevant Focuses on support vector regression rather than language model evaluation.
A Large-Dimensional Analysis of Least-Squares Federated Learning via Generalized Total Variation Minimization The paper uses random matrix theory to analyze the trade-off between noise reduction and bias in decentralized federated learning. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 15:06	Not relevant Covers theoretical federated learning convergence rather than deployment safety or guardrails.
A Large-Dimensional Analysis of Least-Squares Federated Learning via Generalized Total Variation Minimization The paper uses random matrix theory to analyze the trade-off between noise reduction and bias in decentralized federated learning. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 15:06	Not relevant Theoretical study of federated learning, not language model evaluation benchmarks.
A Large-Dimensional Analysis of Least-Squares Federated Learning via Generalized Total Variation Minimization The paper uses random matrix theory to analyze the trade-off between noise reduction and bias in decentralized federated learning. Preprint	Small and efficient models via arxiv · 15 Oct, 15:06	Not relevant Does not focus on model compression, small models, or efficient serving.
What Do Foundation Models for Market Microstructure Learn? A Controlled Study with Synthetic Limit Order Books The paper evaluates autoregressive foundation models trained on simulated market order flows to analyze whether they learn underlying Bayesian dynamics. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 15:06	Not relevant Does not study deployment safety, alignment, hallucinations, or adversarial attacks.
What Do Foundation Models for Market Microstructure Learn? A Controlled Study with Synthetic Limit Order Books The paper evaluates autoregressive foundation models trained on simulated market order flows to analyze whether they learn underlying Bayesian dynamics. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 15:06	Not relevant Investigates market microstructure representation learning rather than LLM evaluation methodology.
What Do Foundation Models for Market Microstructure Learn? A Controlled Study with Synthetic Limit Order Books The paper evaluates autoregressive foundation models trained on simulated market order flows to analyze whether they learn underlying Bayesian dynamics. Preprint	Small and efficient models via arxiv · 15 Oct, 15:06	Not relevant Does not study small language models or inference efficiency techniques.
Exhaustive Canonical Skeletons Generation: A Formal Dataset for Decomposable Symbolic Regression The paper introduces an exhaustive dataset of univariate expression skeletons generated by a formal grammar to improve symbolic regression. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 15:06	Not relevant Focuses on symbolic regression datasets, not language model evaluation or benchmarks.
Exhaustive Canonical Skeletons Generation: A Formal Dataset for Decomposable Symbolic Regression The paper introduces an exhaustive dataset of univariate expression skeletons generated by a formal grammar to improve symbolic regression. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 15:06	Not relevant Not related to safety, hallucinations, jailbreaks, or deployment guardrails.
Exhaustive Canonical Skeletons Generation: A Formal Dataset for Decomposable Symbolic Regression The paper introduces an exhaustive dataset of univariate expression skeletons generated by a formal grammar to improve symbolic regression. Preprint	Small and efficient models via arxiv · 15 Oct, 15:06	Not relevant Does not address model compression, quantization, distillation, or efficient inference.
KANLab: TensorFlow Playground for Kolmogorov-Arnold Networks The paper presents an interactive, browser-based tool for visualizing and learning the mechanics of Kolmogorov-Arnold Networks. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 15:06	Not relevant Interactive educational tool for neural networks, not an evaluation benchmark.
KANLab: TensorFlow Playground for Kolmogorov-Arnold Networks The paper presents an interactive, browser-based tool for visualizing and learning the mechanics of Kolmogorov-Arnold Networks. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 15:06	Not relevant Educational visualization software, not AI deployment safety.
KANLab: TensorFlow Playground for Kolmogorov-Arnold Networks The paper presents an interactive, browser-based tool for visualizing and learning the mechanics of Kolmogorov-Arnold Networks. Preprint	Small and efficient models via arxiv · 15 Oct, 15:06	Not relevant Browser-based visualization tool, not focused on model efficiency.
A Bregman Proximal Viewpoint on Neural Operators The paper introduces a Bregman proximal framework for neural operators to interpret activation functions and improve the training of deeper models. Peer-reviewed	LLM evaluation and benchmarks via arxiv · 15 Oct, 15:06	Not relevant Theoretical paper on neural operators, not an evaluation framework.
A Bregman Proximal Viewpoint on Neural Operators The paper introduces a Bregman proximal framework for neural operators to interpret activation functions and improve the training of deeper models. Peer-reviewed	AI safety and reliability in deployment via arxiv · 15 Oct, 15:06	Not relevant Focuses on mathematical neural operators, not deployment safety.
A Bregman Proximal Viewpoint on Neural Operators The paper introduces a Bregman proximal framework for neural operators to interpret activation functions and improve the training of deeper models. Peer-reviewed	Small and efficient models via arxiv · 15 Oct, 15:06	Not relevant Studies continuous operator learning rather than efficient model inference.
Seeing at the Edge: Energy Characterization of Vision-Language Model Inference on Edge Platforms The paper measures the energy consumption, latency, and thermal behavior of running vision-language models on edge hardware accelerators. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 15:06	Not relevant Examines hardware power characteristics, not deployment safety risks.
Seeing at the Edge: Energy Characterization of Vision-Language Model Inference on Edge Platforms The paper measures the energy consumption, latency, and thermal behavior of running vision-language models on edge hardware accelerators. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 15:06	Not relevant Measures hardware energy profiles, not model quality evaluation benchmarks.
In-Context Batch Black-box Optimization via Flow Matching The paper proposes an amortized flow-matching method to generate diverse batches for parallel black-box optimization over sequential rounds. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 15:06	Not relevant Presents a black-box optimization algorithm, not language model evaluation.
In-Context Batch Black-box Optimization via Flow Matching The paper proposes an amortized flow-matching method to generate diverse batches for parallel black-box optimization over sequential rounds. Preprint	Small and efficient models via arxiv · 15 Oct, 15:06	Not relevant Focuses on batch Bayesian optimization, not efficient language models.
In-Context Batch Black-box Optimization via Flow Matching The paper proposes an amortized flow-matching method to generate diverse batches for parallel black-box optimization over sequential rounds. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 15:06	Not relevant Focuses on black-box optimization methods rather than deployment safety.
Optimizing Sparse Boolean Rule Networks using Mixed Integer Linear Programming The paper presents a mixed integer linear programming formulation to train sparse, interpretable Boolean rule networks with reduced decision variables. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 15:06	Not relevant Focuses on Boolean rule optimization algorithms, not evaluation benchmarks.
Optimizing Sparse Boolean Rule Networks using Mixed Integer Linear Programming The paper presents a mixed integer linear programming formulation to train sparse, interpretable Boolean rule networks with reduced decision variables. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 15:06	Not relevant Focuses on integer programming for rule networks, not safety.
Optimizing Sparse Boolean Rule Networks using Mixed Integer Linear Programming The paper presents a mixed integer linear programming formulation to train sparse, interpretable Boolean rule networks with reduced decision variables. Preprint	Small and efficient models via arxiv · 15 Oct, 15:06	Not relevant Studies small Boolean rule networks, not small language models.
Dimension-Optimal Dynamic Regret in Exp-Concave Online Learning The paper derives dimension-optimal dynamic regret bounds for online learning with exp-concave losses under time-varying comparator paths. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 15:06	Not relevant Presents theoretical online convex optimization regret bounds, not benchmarking.
Dimension-Optimal Dynamic Regret in Exp-Concave Online Learning The paper derives dimension-optimal dynamic regret bounds for online learning with exp-concave losses under time-varying comparator paths. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 15:06	Not relevant Theoretical online learning regret analysis, not AI deployment safety.
Dimension-Optimal Dynamic Regret in Exp-Concave Online Learning The paper derives dimension-optimal dynamic regret bounds for online learning with exp-concave losses under time-varying comparator paths. Preprint	Small and efficient models via arxiv · 15 Oct, 15:06	Not relevant Theoretical optimization theory paper, not focused on efficient models.

6,181–6,240 of 17,313← Previous Next →