Key papers

• Why Multi-Step Tool-Use Reinforcement Learning Collapses and How Supervisory Signals Fix It (2026)
  Yupu Hao et al.
  10.42
• Smarter and Cheaper at Once: Byte-Exact KV-Cache Grafting Turns a Frozen Small Model into a Verified-Knowledge Flywheel (2026)
  Sietse Schelpe
  9.96
• Jet-Long: Efficient Long-Context Extension with Dynamic Bifocal RoPE (2026)
  Haozhan Tang et al.
  9.60
• Neglected Free Lunch from Post-training: Progress Advantage for LLM Agents (2026)
  Changdae Oh et al.
  8.17
• GRPO, Dr. GRPO, and DAPO Are Three Operations on One Number: The Group-Standard-Deviation Identity (2026)
  Yong Yi Bay et al.
  7.37
• CAVEWOMAN: How Large Language Models Behave Under Linguistic Input and Output Compression (2026)
  Morayo Danielle Adeyemi et al.
  6.91
• ReFreeKV: Towards Threshold-Free KV Cache Compression (2025)
  Xuanfan Ni et al.
  6.60
• Is One Layer Enough? Training A Single Transformer Layer Can Match Full-Parameter RL Training (2026)
  Zijian Zhang et al.
  5.87
• LACUNA: A Testbed for Evaluating Localization Precision for LLM Unlearning (2026)
  Matteo Boglioni et al.
  5.49
• Active-GRPO: Adaptive Imitation and Self-Improving Reasoning for Molecular Optimization (2026)
  Xuefeng Liu et al.
  5.01
• The Dual Nature of LLM Persona: Aggregated Tendencies and Frame-Dependent Geometry (2026)
  Yuan Yuan
  5.01
• Q-GAIN: A Python Package for Machine Learning and Physically Informed Analysis Applications (2026)
  M. Doris et al.
  5.01
• Towards Reliable AI-Assisted Analog Design: Template-Constrained LLM Agents for SAR ADC Generation (2026)
  Dimple Vijay Kochar et al.
  5.01
• Surprisal Theory is Tautological (without Rational Grounding) (2026)
  Ryan Cotterell
  5.01
• DWT-Fusion: A Signal-Based Framework for Training-Free LLM-Generated Text Detection (2026)
  Mehmet Batuhan \"Ozda\c{s} et al.
  5.01
• TokenMinds: Pretrained User Tokens and Embeddings for User Understanding in Large Recommender Systems (2026)
  Qingyun Liu et al.
  4.95
• StateFlow: Dual-State Recurrent Modeling for Long-Horizon Time Series Forecasting (2026)
  Haroon Gharwi et al.
  4.39
• Sample Complexities of Estimating Gumbel--Max Watermark Proportions with and without Reduction to Pivotal Statistics (2026)
  Shuwen Chai et al.
  4.39
• Don't Let Gains FADE: Breaking Down Policy Gradient Weights in RL (2026)
  Juliette Decugis et al.
  4.39
• Revisiting Chain-of-Thought Reasoning under Limited Supervision: Semi-supervised Chain-of-Thought Learning (2026)
  Hongyang He et al.
  4.39
• Geometric Signatures of Reasoning: A Spectral Perspective on Task Hardness (2026)
  Aria Masoomi et al.
  4.39
• Epistemic Goggles: A Pretrained Module that Induces an Epistemic Frame via Gradient Editing (2026)
  Joshua Penman
  4.39
• A rubric-based controlled comparison of frontier language models on expert-authored clinical reasoning tasks (2026)
  Samiha A. Ismail et al.
  4.39
• What LLM Agents Say When No One Is Watching: Social Structure and Latent Objective Emergence in Multi-Agent Debates (2026)
  Arman Ghaffarizadeh et al.
  4.39
• Trajectory Inference of Human Aging from Cross-Sectional DNA Methylation Data (2026)
  Chandan Gupta et al.
  4.39
• TriRoute: Unified Learned Routing for Joint Adaptive Attention, Experts, and KV-Cache Allocation (2026)
  Andrii Balashov et al.
  4.39
• LLM-Guided Task-Semantic Field Factorization for Industrial Process Forecasting (2026)
  Youcheng Zong et al.
  4.39
• Efficient Long-Horizon Learning for Learned Optimization (2026)
  Xiaolong Huang et al.
  4.39
• Selective Left-Shift: Turning Test-Time Compute and Difficulty-based Curation into Training Data for Low-Resource Code Generation (2026)
  Didula Samaraweera et al.
  4.39
• When Thinking Hurts: Epistemic Signals in the Reasoning Chains of Visual Language Models (2026)
  Mayank Singal
  4.39
• Super Weights in LLMs and the Failure of Selective Training (2026)
  Shreyas Subramanian et al.
  4.39
• Pitfalls and Remedies for Multi-Task Bayesian Optimization (2026)
  Carl Hvarfner et al.
  4.39
• Branching Policy Optimization: Sandbox-Native Language Agent Reinforcement Learning (2026)
  Bowei He et al.
  4.39
• Non-vacuous Generalization Bounds for Reinforcement Learning with Verifiable Rewards (2026)
  Yuxuan Zhu et al.
  4.39
• ExaGEMM: Exploration Framework for CPU-Driven ML Inference via Associative In-Register Computing for Low-Bit GEMM (2026)
  Hyunwoo Oh et al.
  4.39
• Innocuous-Seeming Data, Latent Ideology: Ideological Generalisation in Finetuned LLMs (2026)
  Robert Graham et al.
  4.39
• Convergence-Latency-Aware Adaptive Modulation and Resource Allocation in RIS-Assisted Wireless Federated Learning (2026)
  Liwei Wang et al.
  4.39
• Autonomous Collaborative Learning Among an Ensemble of Tsetlin Machines with Consensus-Based Inference (2026)
  Yehuda Rudin et al.
  4.39
• Grounding Investor Views: Neural Predicates in the Black-Litterman Model (2026)
  Marcos Florencio
  4.39
• Physically Constrained Federated Additive Models for O-RAN SLA-Risk Prediction (2026)
  Aubida A. Al-Hameed et al.
  4.39
• Searching the Space of Feed-Forward Neural-Network Weight-Update Rules with Fixed Depth Symbolic Regression (2026)
  Charles Brum et al.
  4.39
• A Leakage-Free Stacked Ensemble Method for Multiclass Classification (2026)
  S. P. Sharmila et al.
  4.39
• Zeta: Dual Whitening for Matrix Optimization via Coordinate-Adaptive Preconditioning (2026)
  Kaiwen Chen et al.
  4.33
• Efficiently Representing Algorithms With Chain-of-Thought Transformers (2026)
  Yanhong Li et al.
  4.33
• What Do Safety-Aligned LLMs Learn From Mixed Compliance Demonstrations? (2026)
  Sihui Dai et al.
  4.33
• Exposing the Illusion of Erasure in Knowledge Editing for LLMs (2026)
  Advik Raj Basani et al.
  4.33
• AsyncOPD: How Stale Can On-Policy Distillation Be? (2026)
  Wonjun Kang et al.
  4.33
• Toward Self-Evolution-Ready Workflow Harnesses: A Reversible Migration Path and Convertibility Taxonomy for Expert LLM Pipelines (2026)
  Yimo Lin et al.
  4.33
• Curvature-Guided Mixing for MLLM Adaptation (2026)
  Jinglong Yang et al.
  4.33
• ASAP: Agent-System Co-Design for Wall-Clock-Centered Auto HPO Research for ML Experiments (2026)
  Taicheng Guo et al.
  4.33
• Same Evidence, Different Answer: Auditing Order Sensitivity in Multimodal Large Language Models (2026)
  Akshay Paruchuri et al.
  4.33
• HyperDFlash: Hyper-Connection-Aligned Block Speculative Decoding with Gated Residual Reduction (2026)
  Luxi Lin et al.
  4.33
• DysLexLens: A Low-Resource LLM Framework for Analysing Dyslexic Learners Insights from Online Forums (2026)
  Dana Rezazadegan et al.
  4.33
• The Simulacrum: Decision-Theoretic Pretraining for Near-Optimal Time-Series Forecasting and Inference (2026)
  Pablo Montero-Manso et al.
  4.33
• Signed-Permutation Coordinate Transport for RMSNorm Transformers (2026)
  John Sweeney
  4.33
• Procedural Memory Distillation: Online Reflection for Self-Improving Language Models (2026)
  Ye Liu et al.
  3.51
• Purified OPSD: On-Policy Self-Distillation Without Losing How to Think (2026)
  Zhanming Shen et al.
  3.51
• DemoPSD: Disagreement-Modulated Policy Self-Distillation (2026)
  Yunhe Li et al.
  3.51
• Predicting LLM Safety Before Release by Simulating Deployment (2026)
  Marcus Williams et al.
  3.51
• Selection Shapes the Boundary: A Preregistered Replication of Monotonicity and Label Agreement in Unselected NLI Populations (2026)
  Haram Choi
  3.51