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
Causal Discovery Under Hard Selection Bias: A New Robust Score-Matching Approach This paper proposes a score-matching approach for causal discovery in additive noise models under hard selection bias and truncation. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:18	Not relevant Focuses on statistical causal discovery rather than AI deployment safety or robustness.
Causal Discovery Under Hard Selection Bias: A New Robust Score-Matching Approach This paper proposes a score-matching approach for causal discovery in additive noise models under hard selection bias and truncation. Preprint	Small and efficient models via arxiv · 15 Oct, 06:18	Not relevant Addresses causal discovery under selection bias, not model efficiency or compression.
Failure Recovery via Belief Updates and World Modeling for Long-Horizon Agents This paper introduces a framework combining persistent belief states with implicit world modeling to help long-horizon language agents recover from failures. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:18	Not relevant Focuses on agent planning recovery rather than safety, guardrails, or alignment risks.
Failure Recovery via Belief Updates and World Modeling for Long-Horizon Agents This paper introduces a framework combining persistent belief states with implicit world modeling to help long-horizon language agents recover from failures. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18	Not relevant Presents an agent framework and distillation method rather than a general evaluation benchmark.
Same Function, Different Trajectory: No Standard Optimizer Is Equivariant to Transformer Reparameterization This paper demonstrates that standard optimizers are sensitive to parameter scaling in Transformers and proposes a symmetry-aware gradient adjustment. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:18	Not relevant Concerns training stability and optimization theory, not deployment safety or reliability.
Same Function, Different Trajectory: No Standard Optimizer Is Equivariant to Transformer Reparameterization This paper demonstrates that standard optimizers are sensitive to parameter scaling in Transformers and proposes a symmetry-aware gradient adjustment. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18	Not relevant Focuses on optimizer reparameterization dynamics rather than LLM evaluation or benchmarks.
Same Function, Different Trajectory: No Standard Optimizer Is Equivariant to Transformer Reparameterization This paper demonstrates that standard optimizers are sensitive to parameter scaling in Transformers and proposes a symmetry-aware gradient adjustment. Preprint	Small and efficient models via arxiv · 15 Oct, 06:18	Not relevant Addresses training optimization geometry, not model compression or efficient inference methods.
From Spectral Saliency to Finite Deletion: Evaluating Selective Language Model Edits This paper analyzes prediction errors in spectral saliency when deleting singular components to selectively edit language model capabilities. Preprint	Small and efficient models via arxiv · 15 Oct, 06:18	Not relevant Studies targeted capability deletion rather than model compression or inference efficiency.
From Spectral Saliency to Finite Deletion: Evaluating Selective Language Model Edits This paper analyzes prediction errors in spectral saliency when deleting singular components to selectively edit language model capabilities. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18	Not relevant Evaluates selective weight editing methods rather than assessing general LLM capabilities.
The Effect of Layer Ordering in Hybrid Architectures This paper analyzes how layer ordering and selection affect performance across tasks in hybrid attention model architectures. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18	Not relevant Analyzes architecture design theory, not benchmark creation or evaluation.
The Effect of Layer Ordering in Hybrid Architectures This paper analyzes how layer ordering and selection affect performance across tasks in hybrid attention model architectures. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:18	Not relevant Focuses on hybrid model architecture optimization, not safety or guardrails.
Distance Prediction for Quantum Codes This paper introduces an attention-based neural network to predict code distances for bivariate bicycle quantum low-density parity-check codes. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18	Not relevant Predicts quantum code distances, not language model evaluation.
Distance Prediction for Quantum Codes This paper introduces an attention-based neural network to predict code distances for bivariate bicycle quantum low-density parity-check codes. Preprint	Small and efficient models via arxiv · 15 Oct, 06:18	Not relevant Application to quantum error correction codes, not efficient AI.
Distance Prediction for Quantum Codes This paper introduces an attention-based neural network to predict code distances for bivariate bicycle quantum low-density parity-check codes. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:18	Not relevant Quantum code distance calculation, not AI deployment safety.
Revisiting Error Feedback for Model Quantization This paper formulates a trace-minimization framework to design causal error-feedback coefficients for neural network weight quantization. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18	Not relevant Focuses on model quantization techniques, not LLM evaluation benchmarks.
Revisiting Error Feedback for Model Quantization This paper formulates a trace-minimization framework to design causal error-feedback coefficients for neural network weight quantization. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:18	Not relevant Compression error feedback, not deployment safety, guardrails, or alignment.
Learning a Group Element from Its Action: Galois Correspondences, Optimal ERM, and an Abel-Ruffini Threshold for the Natarajan Dimension This paper analyzes the sample and mistake complexity of learning group elements from actions using Galois correspondences and multiclass dimensions. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18	Not relevant Abstract learning theory and group theory, not LLM evaluation.
Learning a Group Element from Its Action: Galois Correspondences, Optimal ERM, and an Abel-Ruffini Threshold for the Natarajan Dimension This paper analyzes the sample and mistake complexity of learning group elements from actions using Galois correspondences and multiclass dimensions. Preprint	Small and efficient models via arxiv · 15 Oct, 06:18	Not relevant Theoretical sample complexity analysis, not efficient model deployment.
Learning a Group Element from Its Action: Galois Correspondences, Optimal ERM, and an Abel-Ruffini Threshold for the Natarajan Dimension This paper analyzes the sample and mistake complexity of learning group elements from actions using Galois correspondences and multiclass dimensions. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:18	Not relevant Pure statistical learning theory, not deployed AI safety.
Random Probing for Structural Self-Interactions in Graph Neural Networks This paper proposes a message-passing layer using randomized probing of operator diagonals to capture node self-interactions in graph neural networks. Preprint	Small and efficient models via arxiv · 15 Oct, 06:18	Not relevant Proposes a graph message passing layer, not model compression.
Random Probing for Structural Self-Interactions in Graph Neural Networks This paper proposes a message-passing layer using randomized probing of operator diagonals to capture node self-interactions in graph neural networks. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:18	Not relevant Graph neural network architecture, not AI safety or reliability.
Random Probing for Structural Self-Interactions in Graph Neural Networks This paper proposes a message-passing layer using randomized probing of operator diagonals to capture node self-interactions in graph neural networks. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18	Not relevant Deals with graph neural network expressivity, not LLM evaluation.
Reinforcement Learning for Large-Scale Lifelong Multi-Agent Path Finding with Shared Spatial Fields This paper introduces a reinforcement learning navigation policy using shared spatial fields for large-scale lifelong multi-agent path finding. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18	Not relevant Addresses multi-agent path finding, not language model evaluation.
Reinforcement Learning for Large-Scale Lifelong Multi-Agent Path Finding with Shared Spatial Fields This paper introduces a reinforcement learning navigation policy using shared spatial fields for large-scale lifelong multi-agent path finding. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:18	Not relevant Grid navigation pathfinding, not deployed AI safety or guardrails.
Reinforcement Learning for Large-Scale Lifelong Multi-Agent Path Finding with Shared Spatial Fields This paper introduces a reinforcement learning navigation policy using shared spatial fields for large-scale lifelong multi-agent path finding. Preprint	Small and efficient models via arxiv · 15 Oct, 06:18	Not relevant Covers multi-agent path planning, not model compression or efficiency.
Link Both Views: Cross-Image Routing Attention for Dual-Endpoint V2I Beam-Pair Power Prediction This paper proposes a cross-image routing attention mechanism to predict millimeter-wave beam power from dual-endpoint vehicle-to-infrastructure camera views. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:18	Not relevant Wireless channel communication application, not AI safety or reliability.
Link Both Views: Cross-Image Routing Attention for Dual-Endpoint V2I Beam-Pair Power Prediction This paper proposes a cross-image routing attention mechanism to predict millimeter-wave beam power from dual-endpoint vehicle-to-infrastructure camera views. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18	Not relevant Vehicle-to-infrastructure wireless power prediction, not language model evaluation.
Link Both Views: Cross-Image Routing Attention for Dual-Endpoint V2I Beam-Pair Power Prediction This paper proposes a cross-image routing attention mechanism to predict millimeter-wave beam power from dual-endpoint vehicle-to-infrastructure camera views. Preprint	Small and efficient models via arxiv · 15 Oct, 06:18	Not relevant Covers wireless beam prediction, not efficient model architectures.
Hierarchical Time Series Forecasting with Operator-Valued Kernels This paper presents an operator-valued kernel regression framework that enforces aggregation constraints directly in hierarchical time series forecasting. Preprint	Small and efficient models via arxiv · 15 Oct, 06:18	Not relevant Covers kernel methods for forecasting, not model efficiency.
Hierarchical Time Series Forecasting with Operator-Valued Kernels This paper presents an operator-valued kernel regression framework that enforces aggregation constraints directly in hierarchical time series forecasting. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18	Not relevant Focuses on hierarchical time-series methods, not LLM evaluation.
Hierarchical Time Series Forecasting with Operator-Valued Kernels This paper presents an operator-valued kernel regression framework that enforces aggregation constraints directly in hierarchical time series forecasting. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:18	Not relevant Time-series forecasting theory, not AI safety and reliability.
Operator Surrogates This book chapter provides an overview and mathematical introduction to finite-parametric operator surrogates for approximating maps between infinite-dimensional spaces. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:18	Not relevant Theoretical operator approximation, not operational AI system safety.
Operator Surrogates This book chapter provides an overview and mathematical introduction to finite-parametric operator surrogates for approximating maps between infinite-dimensional spaces. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18	Not relevant Mathematical survey of operator surrogates, not LLM evaluation.
Operator Surrogates This book chapter provides an overview and mathematical introduction to finite-parametric operator surrogates for approximating maps between infinite-dimensional spaces. Preprint	Small and efficient models via arxiv · 15 Oct, 06:18	Not relevant Covers infinite-dimensional operator approximation, not efficient language modeling.
When Does Opponent Modeling Pay Off? Spatial Inference Stabilises Reinforcement Learning under Partial Observability This paper introduces a spatial transformer opponent model that stabilizes reinforcement learning training under partial observability in competitive foraging games. Preprint	Small and efficient models via arxiv · 15 Oct, 06:18	Not relevant Focuses on RL opponent models, not model compression.
When Does Opponent Modeling Pay Off? Spatial Inference Stabilises Reinforcement Learning under Partial Observability This paper introduces a spatial transformer opponent model that stabilizes reinforcement learning training under partial observability in competitive foraging games. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18	Not relevant Studies spatial RL opponent modeling, not LLM evaluation benchmarks.
When Does Opponent Modeling Pay Off? Spatial Inference Stabilises Reinforcement Learning under Partial Observability This paper introduces a spatial transformer opponent model that stabilizes reinforcement learning training under partial observability in competitive foraging games. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:18	Not relevant RL training stability in games, not deployment safety or guardrails.
When Does Opponent Modeling Pay Off? Spatial Inference Stabilises Reinforcement Learning under Partial Observability This paper introduces a spatial transformer opponent model that stabilizes reinforcement learning training under partial observability in competitive foraging games. Preprint	AI agents via arxiv · 15 Oct, 06:18	Not relevant Studies game-theoretic reinforcement learning, not LLM-based autonomous planning agents.
Teach Me After You Learn It: Lookahead Self-Distillation for RLVR This paper proposes Lookahead Self-Distillation, providing cross-instance token-level supervision to improve reinforcement learning with verifiable rewards for language models. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:18	Not relevant Focuses on reasoning distillation, not deployment safety, jailbreaks, or guardrails.
Teach Me After You Learn It: Lookahead Self-Distillation for RLVR This paper proposes Lookahead Self-Distillation, providing cross-instance token-level supervision to improve reinforcement learning with verifiable rewards for language models. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18	Not relevant Proposes a training method, not a benchmark or evaluation methodology.
Optimal Online Regret for Context-Lumpable Bandits This paper develops an online algorithm achieving optimal regret bounds for stochastic contextual bandits with unknown context-lumpable cluster structures. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18	Not relevant Theoretical contextual bandit paper, not LLM evaluation.
Optimal Online Regret for Context-Lumpable Bandits This paper develops an online algorithm achieving optimal regret bounds for stochastic contextual bandits with unknown context-lumpable cluster structures. Preprint	Small and efficient models via arxiv · 15 Oct, 06:18	Not relevant Addresses theoretical regret bounds, not small or efficient models.
Optimal Online Regret for Context-Lumpable Bandits This paper develops an online algorithm achieving optimal regret bounds for stochastic contextual bandits with unknown context-lumpable cluster structures. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:18	Not relevant Bandit optimization theory, not operational AI safety or alignment.
PING: physics informed graph neural network for advection-diffusion processes with applications to solar power forecasting This paper introduces a physics-informed graph neural network for modeling advection-diffusion processes to forecast solar power production. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:18	Not relevant Focuses on weather and solar modeling, not AI safety.
PING: physics informed graph neural network for advection-diffusion processes with applications to solar power forecasting This paper introduces a physics-informed graph neural network for modeling advection-diffusion processes to forecast solar power production. Preprint	Small and efficient models via arxiv · 15 Oct, 06:18	Not relevant Covers physics-informed graph neural networks, not model compression.
PING: physics informed graph neural network for advection-diffusion processes with applications to solar power forecasting This paper introduces a physics-informed graph neural network for modeling advection-diffusion processes to forecast solar power production. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18	Not relevant Addresses solar power forecasting, not language model evaluation.
Reliable Risk Estimation with Kernel Mean Matching under Dependent Sampling This paper derives finite-sample risk bounds for Kernel Mean Matching under dependent sampling using effective sample sizes. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18	Not relevant Statistical kernel mean matching theory, not LLM evaluation.
Reliable Risk Estimation with Kernel Mean Matching under Dependent Sampling This paper derives finite-sample risk bounds for Kernel Mean Matching under dependent sampling using effective sample sizes. Preprint	Small and efficient models via arxiv · 15 Oct, 06:18	Not relevant Focuses on statistical covariate shift, not efficient language modeling.
Reliable Risk Estimation with Kernel Mean Matching under Dependent Sampling This paper derives finite-sample risk bounds for Kernel Mean Matching under dependent sampling using effective sample sizes. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:18	Not relevant Concerns theoretical risk estimation bounds, not deployment AI safety.
Bayesian Optimization of CVaR with Unknown Hyperparameters This paper presents HE-CV-UCB, an algorithm for Bayesian optimization of Conditional Value-at-Risk under unknown Gaussian process hyperparameters. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:18	Not relevant Mathematical risk optimization, not LLM safety, guardrails, or hallucination.
Bayesian Optimization of CVaR with Unknown Hyperparameters This paper presents HE-CV-UCB, an algorithm for Bayesian optimization of Conditional Value-at-Risk under unknown Gaussian process hyperparameters. Preprint	Small and efficient models via arxiv · 15 Oct, 06:18	Not relevant Presents Bayesian optimization regret bounds, not model efficiency.
Bayesian Optimization of CVaR with Unknown Hyperparameters This paper presents HE-CV-UCB, an algorithm for Bayesian optimization of Conditional Value-at-Risk under unknown Gaussian process hyperparameters. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18	Not relevant Addresses Bayesian optimization algorithms, not language model evaluation.
Over-parameterization Is Neither Robust nor Fragile in Matrix Sensing: A Geometric Theory of Solution Robustness This paper analyzes the geometric robustness of over-parameterized low-rank matrix sensing under the Burer-Monteiro factorization. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18	Not relevant Studies matrix sensing optimization geometry, not LLM evaluation.
Over-parameterization Is Neither Robust nor Fragile in Matrix Sensing: A Geometric Theory of Solution Robustness This paper analyzes the geometric robustness of over-parameterized low-rank matrix sensing under the Burer-Monteiro factorization. Preprint	Small and efficient models via arxiv · 15 Oct, 06:18	Not relevant Covers theoretical over-parameterization in matrix sensing, not model compression.
Over-parameterization Is Neither Robust nor Fragile in Matrix Sensing: A Geometric Theory of Solution Robustness This paper analyzes the geometric robustness of over-parameterized low-rank matrix sensing under the Burer-Monteiro factorization. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:18	Not relevant Examines mathematical optimization stability, not operational AI system safety.
On the Optimality of Random Features under General Conditions: Misspecified Cases and Robust Learning This paper provides optimal learning rate bounds for random feature kernel ridge regression under misspecified models and heavy-tailed noise. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:18	Not relevant Statistical noise robustness theory, not deployed AI safety or guardrails.
On the Optimality of Random Features under General Conditions: Misspecified Cases and Robust Learning This paper provides optimal learning rate bounds for random feature kernel ridge regression under misspecified models and heavy-tailed noise. Preprint	Small and efficient models via arxiv · 15 Oct, 06:18	Not relevant Deals with kernel random feature theory, not language model efficiency.
On the Optimality of Random Features under General Conditions: Misspecified Cases and Robust Learning This paper provides optimal learning rate bounds for random feature kernel ridge regression under misspecified models and heavy-tailed noise. Preprint	LLM evaluation and benchmarks via arxiv · 15 Oct, 06:18	Not relevant Analyzes theoretical kernel ridge regression bounds, not LLM evaluation.
Beyond Task Success: A Theory of Representation Reuse This paper develops a theoretical framework characterizing when representations learned on a source task remain recoverable and reusable for future tasks. Preprint	AI safety and reliability in deployment via arxiv · 15 Oct, 06:18	Not relevant Theoretical representation study unrelated to AI safety and deployment guardrails.
Beyond Task Success: A Theory of Representation Reuse This paper develops a theoretical framework characterizing when representations learned on a source task remain recoverable and reusable for future tasks. Preprint	Small and efficient models via arxiv · 15 Oct, 06:18	Not relevant Discusses downstream representation reuse, not model compression or efficiency.

9,421–9,480 of 17,313← Previous Next →