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
Universal Bounds for Self-Normalized Kullback-Leibler Estimation Establishes finite-sample bounds for self-normalized Kullback-Leibler divergence estimators and applies them to test-time alignment sampling techniques. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Focuses on theoretical information bounds rather than AI safety or reliability.
Universal Bounds for Self-Normalized Kullback-Leibler Estimation Establishes finite-sample bounds for self-normalized Kullback-Leibler divergence estimators and applies them to test-time alignment sampling techniques. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Provides mathematical bounds for KL estimation rather than LLM benchmarking.
Robust Langevin Sampling under State-Dependent Heavy-Tailed Noise with Applications to LLM Fine-Tuning Analyzes Langevin sampling algorithms under state-dependent heavy-tailed gradient noise with applications to Bayesian fine-tuning of language models. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Focuses on theoretical sampling convergence rather than efficient inference or small models.
Robust Langevin Sampling under State-Dependent Heavy-Tailed Noise with Applications to LLM Fine-Tuning Analyzes Langevin sampling algorithms under state-dependent heavy-tailed gradient noise with applications to Bayesian fine-tuning of language models. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Theoretical sampling analysis without direct focus on deployment safety or guardrails.
Robust Langevin Sampling under State-Dependent Heavy-Tailed Noise with Applications to LLM Fine-Tuning Analyzes Langevin sampling algorithms under state-dependent heavy-tailed gradient noise with applications to Bayesian fine-tuning of language models. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Theoretical Langevin Monte Carlo optimization paper, not LLM benchmark evaluation.
When Solvers Get Moody: Quantifying and Predicting Performance Variability in Mixed Integer Programming Proposes a weighted metric to quantify and predict performance variability across mixed-integer programming solver runs. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Addresses solver performance variance in combinatorial optimization, not AI safety.
When Solvers Get Moody: Quantifying and Predicting Performance Variability in Mixed Integer Programming Proposes a weighted metric to quantify and predict performance variability across mixed-integer programming solver runs. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Focuses on operations research MIP solvers, not small or efficient neural models.
When Solvers Get Moody: Quantifying and Predicting Performance Variability in Mixed Integer Programming Proposes a weighted metric to quantify and predict performance variability across mixed-integer programming solver runs. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Studies classical mixed-integer programming solver variability, not LLM evaluation.
HART: Hybrid Attention-RWKV Time series Foundation Model Introduces a hybrid attention-RWKV architecture to mitigate horizon degradation in non-autoregressive time series forecasting. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Evaluates on time series benchmarks but focuses on a new architecture, not LLM evaluation.
HART: Hybrid Attention-RWKV Time series Foundation Model Introduces a hybrid attention-RWKV architecture to mitigate horizon degradation in non-autoregressive time series forecasting. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Does not investigate AI safety, guardrails, or deployment security.
HART: Hybrid Attention-RWKV Time series Foundation Model Introduces a hybrid attention-RWKV architecture to mitigate horizon degradation in non-autoregressive time series forecasting. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Focuses on time-series foundation model architecture design rather than model compression/efficiency.
CoGameReg: Cooperative-Game-Based Collaborative Semi-Supervised Regression for Seismic Reservoir Property Prediction Presents a cooperative game theoretic semi-supervised regression framework for predicting reservoir properties from seismic data. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Presents a domain-specific regression framework, not small or efficient language models.
CoGameReg: Cooperative-Game-Based Collaborative Semi-Supervised Regression for Seismic Reservoir Property Prediction Presents a cooperative game theoretic semi-supervised regression framework for predicting reservoir properties from seismic data. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Applies semi-supervised regression to seismic property prediction, not LLM benchmarking.
CoGameReg: Cooperative-Game-Based Collaborative Semi-Supervised Regression for Seismic Reservoir Property Prediction Presents a cooperative game theoretic semi-supervised regression framework for predicting reservoir properties from seismic data. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Geophysical prediction method unrelated to AI safety, hallucination, or deployment guardrails.
Personalized Federated Direct Preference Optimization via LoRA Adapters Develops a personalized federated preference optimization framework maintaining local LoRA adapters atop a globally synchronized model. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Uses LoRA for federated personalization, not model distillation, quantization, or efficient serving.
Personalized Federated Direct Preference Optimization via LoRA Adapters Develops a personalized federated preference optimization framework maintaining local LoRA adapters atop a globally synchronized model. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Proposes a federated preference optimization algorithm, not an evaluation framework or benchmark.
Generalized Determinantal Point Processes for Attraction and Repulsion with Distinct Interaction Kernels Introduces determinant-ratio point processes to jointly model attraction and repulsion with distinct symmetric positive-definite kernels. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Theoretical statistics paper on determinantal point processes, not LLM evaluation.
Generalized Determinantal Point Processes for Attraction and Repulsion with Distinct Interaction Kernels Introduces determinant-ratio point processes to jointly model attraction and repulsion with distinct symmetric positive-definite kernels. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Focuses on point process sampling and formulation, not model compression or efficient serving.
Generalized Determinantal Point Processes for Attraction and Repulsion with Distinct Interaction Kernels Introduces determinant-ratio point processes to jointly model attraction and repulsion with distinct symmetric positive-definite kernels. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Does not address AI safety, jailbreaks, hallucination, or deployment reliability.
GIST: Generator-state-Informed Supervision for Tabular Foundation Models Proposes averaging query losses over generator states to improve pretraining sample efficiency in tabular foundation models. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Not focused on AI deployment safety, guardrails, or robustness risks.
GIST: Generator-state-Informed Supervision for Tabular Foundation Models Proposes averaging query losses over generator states to improve pretraining sample efficiency in tabular foundation models. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Focuses on tabular foundation model pretraining loss, not small/efficient model deployment.
GIST: Generator-state-Informed Supervision for Tabular Foundation Models Proposes averaging query losses over generator states to improve pretraining sample efficiency in tabular foundation models. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Proposes a training method for tabular models rather than evaluating or benchmarking LLMs.
Controlled-Minimax: Stabilizing Adversarial Optimization via Control Theory Applies continuous-time control theory to stabilize adversarial minimax optimization against rotational divergence. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Addresses GAN optimization stability rather than deployed AI safety, guardrails, or reliability.
Controlled-Minimax: Stabilizing Adversarial Optimization via Control Theory Applies continuous-time control theory to stabilize adversarial minimax optimization against rotational divergence. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Focuses on GAN minimax optimization dynamics, not LLM evaluation methods.
Controlled-Minimax: Stabilizing Adversarial Optimization via Control Theory Applies continuous-time control theory to stabilize adversarial minimax optimization against rotational divergence. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Studies continuous-time GAN training stability, not small models or inference efficiency.
Beyond Second-Order Information: Paid Logistic Regression with Noisy Features Analyzes online binary logistic regression where features are observed noisily subject to variable acquisition costs. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Covers feature acquisition costs in online learning, not efficient or small model inference.
Beyond Second-Order Information: Paid Logistic Regression with Noisy Features Analyzes online binary logistic regression where features are observed noisily subject to variable acquisition costs. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Theoretical online logistic regression paper, unrelated to LLM benchmarking or evaluation.
Beyond Second-Order Information: Paid Logistic Regression with Noisy Features Analyzes online binary logistic regression where features are observed noisily subject to variable acquisition costs. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Focuses on noisy bandit-style online regression, not AI deployment safety or robustness.
Sparse Dictionary Learning for Measure-Valued Data via Linearized Optimal Transport: Efficient Algorithms and Sample Complexity Studies dictionary learning on probability measures via linearized optimal transport embeddings and sparse autoencoders. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Presents theoretical dictionary learning via optimal transport, not LLM benchmark evaluation.
Sparse Dictionary Learning for Measure-Valued Data via Linearized Optimal Transport: Efficient Algorithms and Sample Complexity Studies dictionary learning on probability measures via linearized optimal transport embeddings and sparse autoencoders. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Not related to AI safety, red teaming, jailbreaks, or deployment reliability.
Sparse Dictionary Learning for Measure-Valued Data via Linearized Optimal Transport: Efficient Algorithms and Sample Complexity Studies dictionary learning on probability measures via linearized optimal transport embeddings and sparse autoencoders. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Focuses on mathematical measure embeddings rather than small or efficient language model serving.
From Imbalance to Evidence: Deep Nonparametric Testing for Complex Data under Severe Sample Imbalance Proposes an imbalance-aware deep nonparametric two-sample test that leverages majority observations through repeated testing evidence. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Addresses general distribution testing rather than deployment safety, guardrails, or hallucination.
From Imbalance to Evidence: Deep Nonparametric Testing for Complex Data under Severe Sample Imbalance Proposes an imbalance-aware deep nonparametric two-sample test that leverages majority observations through repeated testing evidence. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Does not study model compression, small models, or efficient LLM inference.
From Imbalance to Evidence: Deep Nonparametric Testing for Complex Data under Severe Sample Imbalance Proposes an imbalance-aware deep nonparametric two-sample test that leverages majority observations through repeated testing evidence. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Focuses on theoretical statistical two-sample testing, not LLM evaluation or benchmarks.
DataCupid: Stable Matching Repair for Training Data Acquisition with Delayed Feedback Develops a matching repair framework with theoretical performance guarantees for training data acquisition under delayed feedback. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Focuses on data acquisition matching repair, not AI safety or deployment guardrails.
DataCupid: Stable Matching Repair for Training Data Acquisition with Delayed Feedback Develops a matching repair framework with theoretical performance guarantees for training data acquisition under delayed feedback. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Algorithmic matching repair under delayed feedback, not model compression or efficiency.
DataCupid: Stable Matching Repair for Training Data Acquisition with Delayed Feedback Develops a matching repair framework with theoretical performance guarantees for training data acquisition under delayed feedback. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Theoretical study on stable matching repair for data acquisition, not LLM benchmarks.
Minions: A Dynamic and Distributed Virtual GPU Cloud Built on Idle Apple Silicon Devices Presents a distributed system architecture aggregating idle Apple Silicon devices into a virtual GPU cloud for inference. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant System paper on distributed GPU inference orchestration, not LLM evaluation.
Minions: A Dynamic and Distributed Virtual GPU Cloud Built on Idle Apple Silicon Devices Presents a distributed system architecture aggregating idle Apple Silicon devices into a virtual GPU cloud for inference. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Addresses distributed cluster resource management, not AI safety or reliability.
Diffusion Precipitation Generation with Medium-range Weather Forecasting Employs a conditional diffusion model to generate observation-aligned precipitation forecasts from frozen weather foundation models. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Proposes a diffusion model for precipitation forecasting, not LLM evaluation.
Diffusion Precipitation Generation with Medium-range Weather Forecasting Employs a conditional diffusion model to generate observation-aligned precipitation forecasts from frozen weather foundation models. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Focuses on weather diffusion models, not small language models or efficient serving.
Diffusion Precipitation Generation with Medium-range Weather Forecasting Employs a conditional diffusion model to generate observation-aligned precipitation forecasts from frozen weather foundation models. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Precipitation forecasting generation method, unrelated to AI safety or guardrails.
NucleoNet: A Nuclear-Aware Determinant Ansatz for Multi-Geometry Neural Variational Monte Carlo Introduces a nuclear-aware determinant ansatz for multi-geometry neural-network variational Monte Carlo calculations. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Focuses on variational Monte Carlo in quantum chemistry, not AI safety.
NucleoNet: A Nuclear-Aware Determinant Ansatz for Multi-Geometry Neural Variational Monte Carlo Introduces a nuclear-aware determinant ansatz for multi-geometry neural-network variational Monte Carlo calculations. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Specialized architecture for molecular electronic structure, not general model compression.
NucleoNet: A Nuclear-Aware Determinant Ansatz for Multi-Geometry Neural Variational Monte Carlo Introduces a nuclear-aware determinant ansatz for multi-geometry neural-network variational Monte Carlo calculations. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Presents a quantum chemistry neural wavefunction ansatz, not LLM benchmarking.
Scan-Parallel Variational Inference for event-driven Latent Jump Models Proposes scan-parallel variational inference for continuous-time latent jump models with state-dependent event intensities. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Introduces variational inference for latent jump models, not LLM evaluation.
Scan-Parallel Variational Inference for event-driven Latent Jump Models Proposes scan-parallel variational inference for continuous-time latent jump models with state-dependent event intensities. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Deals with continuous-time state-space models and inference, not language model efficiency.
Scan-Parallel Variational Inference for event-driven Latent Jump Models Proposes scan-parallel variational inference for continuous-time latent jump models with state-dependent event intensities. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Focuses on latent jump models for time series, unrelated to AI safety.
Gaussian-Width Bounds for Offline-to-Online Linear Bandits Derives tighter regret bounds and algorithms for offline-to-online linear bandits using Gaussian concentration tools. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Theoretical analysis of linear bandits, unrelated to small language models or efficient serving.
Gaussian-Width Bounds for Offline-to-Online Linear Bandits Derives tighter regret bounds and algorithms for offline-to-online linear bandits using Gaussian concentration tools. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Theoretical regret bounds for stochastic linear bandits, not LLM benchmarking.
Gaussian-Width Bounds for Offline-to-Online Linear Bandits Derives tighter regret bounds and algorithms for offline-to-online linear bandits using Gaussian concentration tools. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Focuses on bandit theoretical regret bounds, not AI safety or system reliability.
Evidential Multimodal Transformers with Dempster-Shafer Fusion for Emotion Recognition Introduces an evidential transformer using Dempster-Shafer fusion to quantify and handle epistemic and aleatoric multimodal emotion uncertainty. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Presents an emotion recognition model with uncertainty quantification, not LLM benchmarks.
Evidential Multimodal Transformers with Dempster-Shafer Fusion for Emotion Recognition Introduces an evidential transformer using Dempster-Shafer fusion to quantify and handle epistemic and aleatoric multimodal emotion uncertainty. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Deals with multimodal emotion classification under sensor conflict, not LLM deployment safety.
Evidential Multimodal Transformers with Dempster-Shafer Fusion for Emotion Recognition Introduces an evidential transformer using Dempster-Shafer fusion to quantify and handle epistemic and aleatoric multimodal emotion uncertainty. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Focuses on evidential transformers for emotion recognition, not model compression.
Trajectory-Aware Policy Optimization: Closing the State Distribution Gap in Long-Horizon LLM RL Presents a policy optimization method that corrects state-distribution shift during long-horizon reinforcement learning with language models. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Addresses RL policy optimization for reasoning tasks, not safety guardrails or red-teaming.
Trajectory-Aware Policy Optimization: Closing the State Distribution Gap in Long-Horizon LLM RL Presents a policy optimization method that corrects state-distribution shift during long-horizon reinforcement learning with language models. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Focuses on RL policy optimization training dynamics, not model compression or serving.
Trajectory-Aware Policy Optimization: Closing the State Distribution Gap in Long-Horizon LLM RL Presents a policy optimization method that corrects state-distribution shift during long-horizon reinforcement learning with language models. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Proposes an RL training algorithm for LLM reasoning, not an evaluation framework.
Low-Rank Kolmogorov-Arnold Generator Flows for Interpretable Operator Learning Introduces Kolmogorov-Arnold flow to learn physical operator generators using low-rank univariate function symbols. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:29	Not relevant Studies physical operator learning, not AI safety or deployment reliability.
Low-Rank Kolmogorov-Arnold Generator Flows for Interpretable Operator Learning Introduces Kolmogorov-Arnold flow to learn physical operator generators using low-rank univariate function symbols. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Focuses on physics operator learning using Kolmogorov-Arnold flows, not small language models.
Low-Rank Kolmogorov-Arnold Generator Flows for Interpretable Operator Learning Introduces Kolmogorov-Arnold flow to learn physical operator generators using low-rank univariate function symbols. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:29	Not relevant Develops neural operators for physical laws, not LLM benchmarking or evaluation.

15,001–15,060 of 17,313← Previous Next →