Key papers

• ISO: An RLVR-Native Optimization Stack (2026)
  Hanqing Zhu et al.
  8.80
• H$^2$SD: Hybrid Hindsight Self-Distillation (2026)
  Qiye Cai et al.
  8.51
• NexForge: Scaling Agent Capabilities through Requirement-Driven Task Synthesis for LLMs (2026)
  Jiarong Zhao et al.
  7.90
• A Controlled Study of Attention-Only Transformers (2026)
  Henry Ndubuaku et al.
  5.90
• Measuring Reward-Seeking via Contrastive Belief Updates (2026)
  Axel H{\o}jmark et al.
  4.95
• How Much of the Routing Gap Is Real? Decomposing the Router-to-Oracle Gap into Reproducible Specialist Advantage and Single-Draw Label Noise (2026)
  Teng-Ruei Chen
  4.33
• Calibration-Family Overfit: Why Trusted Sabotage Monitors Don't Transfer Across Lineages (2026)
  Lucas Pinto
  4.33
• TriRoute: Unified Learned Routing for Joint Adaptive Attention, Experts, and KV-Cache Allocation (2026)
  Andrii Balashov et al.
  4.33
• HiFuzz: Hierarchical Reinforcement Learning for Semantic-Aware and Adaptive CPU Fuzzing (2026)
  Ya Wang et al.
  4.33
• Open-Ended Scenario Reasoning for Specialist Model Adaptation (2026)
  Youcheng Zong et al.
  4.33
• At-Grok Is Not Converged:A Measurement-Validity Audit for Grokking Representation Metrics (2026)
  Truong Xuan Khanh
  4.33
• The Rank-One Corner: How Much Value Equivalence Does a Task Need from a World Model? (2026)
  Donna Vakalis
  4.33
• Can a Language Model Learn Facts Continually in Its Weights? (2026)
  Charles O'Neill
  4.33
• Bringing Back Rule Induction to Fluid Intelligence Research? An Initial Validation of the ARC-AGI Benchmark in Humans (2026)
  Jasmin Thelen et al.
  4.33
• AAAI-26 Dual Submissions: Novel Challenges (2026)
  Kiri L. Wagstaff et al.
  4.33
• CARE-LoRA: Compressed Activation REconstruction for Memory-Efficient LoRA (2026)
  Gengyu Zhang et al.
  4.33
• When Does Reward Teach State? A Hidden-Automaton Instrument and the Group-Language Boundary (2026)
  Jim Allchin
  4.33
• Constructed Reality, Contested Priors: Decoupling and the Architecture of Cognitive Relapse Under the Free Energy Principle (2026)
  MD Ibrahim Hossain Ridoy
  4.33
• Removable Defects: The Economics and Limits of Deliberate Deficiency (2026)
  Cheng Qian
  4.33
• Are we Merging the Right Models? Impact of Expert Training Duration on Model Merging for LLMs (2026)
  Nikita Kozodoi et al.
  4.33
• Continual Learning with Elastic Regularization and Synthetic Replay for Federated MLLM Fine-Tuning (2026)
  Jing Liu et al.
  4.33
• Forgetful Attention: A Trainable Support-Vector Memory with Certified Selection and Exact Unlearning (2026)
  Vishwajith Ramesh
  4.33
• Speculate with Memory: Lossless Acceleration for LLM Agents (2026)
  Yu Li et al.
  4.33
• The Sound of Absence: Audio-Language Embedding Models Struggle with Negation (2026)
  Chun-Yi Kuan et al.
  4.33
• Extractable Memorization From First Principles (2026)
  A. Feder Cooper et al.
  4.33
• What Makes a Representational Prior Work? Feature Families, Label-Free Invariances, and Critical Windows in Grokking (2026)
  Gunner Levi Howe
  4.33
• Quantifying Diversity of Thought: A Predictive Law of Weighted LLM Ensemble Lift (2026)
  Junade Ali
  4.33
• Can AI Agents Really Complete RTL-to-GDS? Lessons from Benchmarking Tool-Interactive EDA Workflows (2026)
  Jinyuan Deng et al.
  4.33
• Beyond Single-Dimensional Compression: The Compound Sparsity Frontier of Large Language Models (2026)
  Chao Han et al.
  4.33
• One Student, Many Teachers: Multi-Task On-Policy Distillation via Soft-Prompt Privileged Context (2026)
  Yingzi Ma et al.
  4.33
• On the Limits of Support-Preserving Alignment and Bounded Filtering (2026)
  Aryan Dutt et al.
  4.33
• A Better Start for Language Models: Domain-Conditional Position Offsets (2026)
  Ye Qiao
  4.33
• TD-DPO: Difference-Aware Preference Optimization for Mitigating Sycophancy in Clinical Autism Intervention Dialogue (2026)
  Shuzhong Lai et al.
  4.33
• The Information Shadow: Measuring Structural Limits on What Language Models Can Learn (2026)
  Priyansh Srivastava et al.
  4.33
• Agentic Calibration of Grey-Box Simulation Models: An LLM-Driven Alternative (2026)
  David G\'omez-Guill\'en et al.
  4.33
• Estimating Rare Events in Language Models with Proper Evaluation (2026)
  Nikita Y. Parulekar et al.
  4.33
• Operational Proto-Introspection in Looped Language Models: Process-Quality Taps, Executable Branching, and the Readout-Control Boundary (2026)
  Jan Kirin
  4.33
• Attacking Graph Foundation Models Through Their Shared Representation (2026)
  Pankaj Kumar et al.
  4.33
• BRIDGE: Bottleneck-Aware Regulator-Set Inference and Diagnosis for Cooperative Gene Regulatory Recovery (2026)
  Maryam Rahimimovassagh et al.
  4.33
• Mark, Don't Erase: Token Inoculation for Dual-Use Knowledge in LLMs (2026)
  Seunghyun Lee et al.
  4.33
• Spaghetti Architect: A Contamination-Resistant, By-Construction-Labelled, Multi-Language Code Dataset Generator (2026)
  Yuxiang Ji
  4.33
• Exposure-Based Reinforcement Learning to Rank (2026)
  Harrie Oosterhuis et al.
  4.33
• Relative Positions Generalize, Absolute Positions Memorize: An Implicit-Bias Account of Length Generalization in Attention (2026)
  Subham Singh et al.
  4.33
• PertReason: A Knowledge-Grounded Benchmark and Framework for Cell-State-Conditioned Mechanistic Reasoning of Perturbation Effects (2026)
  Dongkwan Kim et al.
  4.33
• Elicitation without Backpropagation: Steering Model Behavior by Optimizing the Latent Posterior (2026)
  Garrett Baker et al.
  4.33
• From Trajectories to Instructions: Language-Conditioned Meta-Reinforcement Learning (2026)
  Garvit Singla et al.
  4.33
• ABOPD: Antibody CDR Design via On-Policy Distillation (2026)
  Zhuo Yang et al.
  4.33
• HindsightBench: A Black-Box Behavioral Audit Protocol for Parametric Hindsight in Time-Indexed LLM Decision Tasks (2026)
  Haozhe Jia
  4.33
• Breaking Feedback-Blindness: Utility-Augmented Transformer for Sequential Decision Making (2026)
  Yuyang Shen et al.
  4.33
• Circuit Claims Depend on What Is Extracted and How It Is Compared (2026)
  Yang Sheng et al.
  4.33
• Adopting Reinforcement Learning with Verifiable Rewards for Molecular Generation (2026)
  Mingxuan Ouyang et al.
  4.33
• Parallel Noising in Neural Markov Logic Networks (2026)
  Peter Jung et al.
  4.33
• AdaFlash: Adaptive Speculative Decoding via On-Policy Distilled Diffusion Drafters (2026)
  Yu-Yang Qian et al.
  4.33
• Benchmarking Generalization in Financial Statement Fraud Detection: robust evaluation and novel tasks (2026)
  Guy Stephane Waffo Dzuyo (Forvis Mazars et al.
  4.33
• Off-Context GRPO: Learning to Reason on Hard Problems using Privileged Information (2026)
  Priyank Agrawal et al.
  4.33
• Reading Without a Reader: Large Language Models Collapse Reading and Writing into a Single Entangled Code (2026)
  Diego Salda\~na Ulloa
  4.33
• ESC: Emotional Self-Correction for Reliable Vision-Language Models (2026)
  Tien-Huy Nguyen et al.
  3.45
• Rank-Order N-of-M Codes for Sparse Distributed Memory: Disentangling Representation and Learning Effects in Noise Robustness Against Contemporary Neuromorphic Architectures (2026)
  Joy Bose
  3.45
• ACPO: Adaptive Credit Policy Optimization via Fine-Grained Surrogate Entropy (2026)
  Zijun Xie et al.
  3.45
• To Retain or to Adapt? Generalizing Continual Learning (2026)
  Giulia Lanzillotta et al.
  3.45