| Domino: Decoupling Causal Modeling from Autoregressive Drafting in Speculative Decoding | Code | Jianuo Huang +5 | May 28, 2026 | 143 |
| Your UnEmbedding Matrix is Secretly a Feature Lens for Text Embeddings | Code | Songhao Wu +5 | Jun 5, 2026 | 91 |
| ProRL: Effective Reinforcement Learning for Proactive Recommendation via Rectified Policy Gradient Estimation | Code | Hongru Hou +7 | May 27, 2026 | 87 |
| Code2LoRA: Hypernetwork-Generated Adapters for Code Language Models under Software Evolution | Code | Liliana Hotsko +3 | Jun 4, 2026 | 86 |
| FlashMemory-DeepSeek-V4: Lightning Index Ultra-Long Context via Lookahead Sparse Attention | Code | Yan Wang +12 | Jun 8, 2026 | 61 |
| KVarN: Variance-Normalized KV-Cache Quantization Mitigates Error Accumulation in Reasoning Tasks | Code | Lorenz K. Muller +5 | Jun 2, 2026 | 59 |
| TRL-Bench: Standardizing Cross-Paradigm Representation-Level Evaluation of Tabular Encoders | Code | Wei Pang +12 | Jun 8, 2026 | 50 |
| How LoRA Remembers? A Parametric Memory Law for LLM Finetuning | Code | Ziwen Xu +6 | May 28, 2026 | 42 |
| ThriftAttention: Selective Mixed Precision for Long-Context FP4 Attention | Code | Joe Sharratt | May 21, 2026 | 41 |
| MemTrace: Tracing and Attributing Errors in Large Language Model Memory Systems | Code | Xinle Deng +17 | May 27, 2026 | 39 |
| Draft-OPD: On-Policy Distillation for Speculative Draft Models | Code | Haodi Lei +10 | May 28, 2026 | 33 |
| VIA-SD: Verification via Intra-Model Routing for Speculative Decoding | Code | Yuchen Xian +3 | Jun 10, 2026 | 32 |
| A Local Perturbation Theory for Cross-Domain Interference and Recovery in Multi-Domain RL | Code | Lei Yang +2 | Jun 1, 2026 | 27 |
| Reinforcement Learning Elicits Contextual Learning of Unseen Language Translation | Code | Hanxu Hu +4 | Jun 4, 2026 | 25 |
| Self-Distilled Policy Gradient | Code | Yifeng Liu +3 | Jun 2, 2026 | 24 |
| When Should Models Change Their Minds? Contextual Belief Management in Large Language Models | Code | Haoming Xu +8 | May 28, 2026 | 24 |
| On Subquadratic Architectures: From Applications to Principles | Code | Anamaria-Roberta Hartl +8 | Jun 10, 2026 | 23 |
| Long Live The Balance: Information Bottleneck Driven Tree-based Policy Optimization | Code | Hao Jiang +9 | May 27, 2026 | 23 |
| MIRA: Mid-training Rubric Anchoring for Source-Aware Data Selection | Code | Haowen Wang +10 | May 28, 2026 | 22 |
| Xetrieval: Mechanistically Explaining Dense Retrieval | Code | Zhixin Cai +9 | May 28, 2026 | 21 |
| Domain-Specific Data Synthesis for LLMs via Minimal Sufficient Representation Learning | Code | Tong Ye +8 | May 28, 2026 | 18 |
| Grammar-Constrained Decoding Can Jailbreak LLMs into Generating Malicious Code | Code | Yitong Zhang +2 | Jun 10, 2026 | 17 |
| Filter, Then Reweight: Rethinking Optimization Granularity in On-Policy Distillation | Code | Yuying Li +8 | Jun 1, 2026 | 16 |
| HRBench: Benchmarking and Understanding Thinking-Mode Switch Strategies in Hybrid-Reasoning LLMs | Code | Yansong Ning +4 | May 27, 2026 | 15 |
| DEI: Diversity in Evolutionary Inference for Quality-Diversity Search | Code | John Donaghy +1 | May 26, 2026 | 14 |
| CubePart: An Open-Vocabulary Part-Controllable 3D Generator | Code | Yiheng Zhu +11 | May 27, 2026 | 14 |
| Towards Evaluation Engineering: An Empirical Study of ML Evaluation Harnesses in the Wild | Code | Zhimin Zhao +4 | May 22, 2026 | 14 |
| Lost in Sampling: Assessing Lexical Reachability in LLMs via the Word Coverage Score (WCS) | Code | Samer Awad +5 | May 26, 2026 | 13 |
| Confidence and Calibration of Activation Oracles for Reliable Interpretation of Language Model Internals | Code | Federico Torrielli +2 | May 25, 2026 | 12 |
| Speculative Pipeline Decoding: Higher-Accruacy and Zero-Bubble Speculation via Pipeline Parallelism | Code | Yijiong Yu +4 | May 29, 2026 | 10 |
| Verifiable Rewards Beyond Math and Code: Lightweight Corpus-Grounded Process Supervision for Factual Question Answering | Code | Shicheng Fan +5 | May 28, 2026 | 10 |
| Thinking Before Constraining: A Unified Decoding Framework for Large Language Models | Code | Ngoc Trinh Hung Nguyen +5 | May 28, 2026 | 10 |
| OPRD: On-Policy Representation Distillation | Code | Shenzhi Yang +9 | Jun 4, 2026 | 9 |
| Reasoning Arena: Trace Tournaments When Verifiable Rewards Fall Short | Code | Han Zhou +4 | Jun 8, 2026 | 8 |
| Answer Presence Drives RAG Rewriting Gains | Code | Yuejie Li +10 | Jun 4, 2026 | 8 |
| Measuring the Depth of LLM Unlearning via Activation Patching | Code | Jaeung Lee +2 | May 23, 2026 | 8 |
| Models That Know How Evaluations Are Designed Score Safer | Code | Katharina Deckenbach +3 | May 27, 2026 | 8 |
| ZeroUnlearn: Few-Shot Knowledge Unlearning in Large Language Models | Code | Yujie Lin +4 | May 16, 2026 | 8 |
| Trajectory-Refined Distillation | Code | Li Jiang +3 | Jun 7, 2026 | 7 |
| Latent Reasoning with Normalizing Flows | Code | Guancheng Tu +7 | Jun 4, 2026 | 7 |
| LongAttnComp: Cross-Family Context Compression for Long-Context Reasoning | Code | Mengmeng Ji +3 | May 31, 2026 | 7 |
| MAAT: Multi-phase Adapter-Aware Targeted Unlearning | Code | Suryash Yagnik +5 | May 28, 2026 | 7 |
| AlphaTransit: Learning to Design City-scale Transit Routes | Code | Bibek Poudel +2 | May 27, 2026 | 7 |
| Multi-view Consistent 3D Gaussian Head Avatars 'without' Multi-view Generation | Code | Aviral Chharia +1 | May 24, 2026 | 7 |
| DRIFT: Decoupled Rollouts and Importance-Weighted Fine-Tuning for Efficient Multi-Turn Optimization | Code | Jian Mu +4 | May 29, 2026 | 6 |
| The Shape of Addition: Geometric Structures of Arithmetic in Large Language Models | Code | Liuyuan Wen +4 | May 29, 2026 | 5 |
| MeshWeaver: Sparse-Voxel-Guided Surface Weaving for Autoregressive Mesh Generation | Code | Jiale Xu +2 | Jun 3, 2026 | 5 |
| BenchEvolver: Frontier Task Synthesis via Solution-Centric Evolution | Code | Yangzhen Wu +12 | May 31, 2026 | 5 |
| Efficient and Scalable Provenance Tracking for LLM-Generated Code Snippets | Code | Andrea Gurioli +4 | May 27, 2026 | 5 |
| Understanding Data Temporality Impact on Large Language Models Pre-training | Code | Hippolyte Pilchen +4 | May 21, 2026 | 5 |
| U-TTT: Towards Generalizable PET Image Denoising via Test-Time Training | Code | Zhiwen Yang +6 | Jun 9, 2026 | 4 |
| IR3DE: A Linear Router for Large Language Models | Code | Eros Fanì +1 | Jun 4, 2026 | 4 |
| Unified Panoramic Geometry Estimation via Multi-View Foundation Models | Code | Vukasin Bozic +6 | May 25, 2026 | 4 |
| NSF-SciFy: Mining the NSF Awards Database for Scientific Claims | Code | Delip Rao +3 | May 25, 2026 | 4 |
| When is Your LLM Steerable? | Code | Chenrui Fan +4 | Jun 10, 2026 | 3 |
| SparDA: Sparse Decoupled Attention for Efficient Long-Context LLM Inference | Code | Yaosheng Fu +4 | Jun 3, 2026 | 3 |
| Reinforcement Learning from Rich Feedback with Distributional DAgger | Code | Rishabh Agrawal +2 | Jun 3, 2026 | 3 |
| Surflo: Consistent 3D Surface Flow Model with Global State | Code | Antoine Guédon +5 | Jun 11, 2026 | 2 |
| Adaptive Multi-Resolution Procedural Knowledge Compression for Large Language Models | Code | Changyue Wang +7 | Jun 10, 2026 | 2 |
| Combinatorial Synthesis: Scaling Code RLVR via Atomic Decomposition and Recombination | Code | Jiasheng Zheng +8 | May 29, 2026 | 2 |
| BA-T: An Iterative Transformer for Two-View Bundle Adjustment | Code | Ganlin Zhang +3 | Jun 2, 2026 | 2 |
| Confidence-Adaptive SwiGLU for Mixture-of-Experts | Code | Shaohua Li +6 | May 30, 2026 | 2 |
| A Stationary (and Therefore Compatible) Representation is All You Need | Code | Niccolò Biondi +3 | Jun 10, 2026 | 1 |
| Empirical Study on the Characteristics and Evolution of AI-usage in GitHub Repositories: Evidence from Code Comments | Code | Abdullah Al Mujahid +2 | Jun 5, 2026 | 1 |
| Measuring Model Robustness via Fisher Information: Spectral Bounds, Theoretical Guarantees, and Practical Algorithms | Code | Chong Zhang +4 | Jun 3, 2026 | 1 |
| The Distillation Game: Adaptive Attacks & Efficient Defenses | Code | Youssef Allouah +3 | May 21, 2026 | 1 |
| Review Arcade: On the Human Alignment and Gameability of LLM Reviews | Code | Hans Ole Hatzel +2 | May 27, 2026 | 1 |
| BatteryMFormer: Multi-level Learning for Battery Degradation Trajectory Forecasting | Code | Ruifeng Tan +5 | May 26, 2026 | 1 |