{"total":127,"rows":[{"id":"2510.06477","title":"Attention Sinks and Compression Valleys in LLMs are Two Sides of the Same Coin","created":"2025-10-07","cited_by_count":52,"area":null},{"id":"2504.15364","title":"KeyDiff: Key Similarity-Based KV Cache Eviction for Long-Context LLM Inference in Resource-Constrained Environments","created":"2025-04-21","cited_by_count":51,"area":"inference"},{"id":"2604.11791","title":"A Mechanistic Analysis of Looped Reasoning Language Models","created":"2026-04-13","cited_by_count":33,"area":"lr"},{"id":"2505.11695","title":"Qronos: Correcting the Past by Shaping the Future... in Post-Training Quantization","created":"2025-05-16","cited_by_count":28,"area":null},{"id":"2508.02546","title":"What are you sinking? A geometric approach on attention sink","created":"2025-08-04","cited_by_count":28,"area":null},{"id":"2510.13554","title":"Attention Illuminates LLM Reasoning: The Preplan-and-Anchor Rhythm Enables Fine-Grained Policy Optimization","created":"2025-10-15","cited_by_count":25,"area":"lr"},{"id":"2504.20966","title":"Softpick: No Attention Sink, No Massive Activations with Rectified Softmax","created":"2025-04-29","cited_by_count":22,"area":null},{"id":"2510.08510","title":"To Sink or Not to Sink: Visual Information Pathways in Large Vision-Language Models","created":"2025-10-09","cited_by_count":20,"area":"mllm"},{"id":"2510.15731","title":"Attention Sinks in Diffusion Language Models","created":"2025-10-17","cited_by_count":20,"area":"diffusion"},{"id":"2504.07863","title":"Robust Hallucination Detection in LLMs via Adaptive Token Selection","created":"2025-04-10","cited_by_count":19,"area":null},{"id":"2504.10063","title":"Hallucination Detection in LLMs with Topological Divergence on Attention Graphs","created":"2025-04-14","cited_by_count":19,"area":null},{"id":"2609.01343","title":"SMELT: Scaling Laws for Compute-Matched MoE Looped Transformers","created":"2026-09-01","cited_by_count":18,"area":"moe"},{"id":"2506.22058","title":"Lost at the Beginning of Reasoning","created":"2025-06-27","cited_by_count":15,"area":"lr"},{"id":"2506.19697","title":"Outlier-Safe Pre-Training for Robust 4-Bit Quantization of Large Language Models","created":"2025-06-24","cited_by_count":13,"area":null},{"id":"2505.11739","title":"ZeroTuning: Unlocking the Initial Token's Power to Enhance Large Language Models Without Training","created":"2025-05-16","cited_by_count":12,"area":null},{"id":"2505.18658","title":"Robustness in Large Language Models: A Survey of Mitigation Strategies and Evaluation Metrics","created":"2025-05-24","cited_by_count":12,"area":null},{"id":"2506.07624","title":"Return of ChebNet: Understanding and Improving an Overlooked GNN on Long Range Tasks","created":"2025-06-09","cited_by_count":12,"area":null},{"id":"2507.07814","title":"Pay Attention to Attention Distribution: A New Local Lipschitz Bound for Transformers","created":"2025-07-10","cited_by_count":11,"area":null},{"id":"2509.17178","title":"Attention Consistency for LLMs Explanation","created":"2025-09-21","cited_by_count":11,"area":null},{"id":"2510.22603","title":"Mitigating Attention Sinks and Massive Activations in Audio-Visual Speech Recognition with LLMs","created":"2025-10-26","cited_by_count":11,"area":null},{"id":"2603.11487","title":"Attention Sinks Are Provably Necessary in Softmax Transformers: Evidence from Trigger-Conditional Tasks","created":"2026-03-12","cited_by_count":11,"area":null},{"id":"2508.20412","title":"MindGuard: Intrinsic Decision Inspection for Securing LLM Agents Against Metadata Poisoning","created":"2025-08-28","cited_by_count":10,"area":"agents"},{"id":"2510.04577","title":"Language Model Based Text-to-Audio Generation: Anti-Causally Aligned Collaborative Residual Transformers","created":"2025-10-06","cited_by_count":8,"area":null},{"id":"2510.09017","title":"Value-State Gated Attention for Mitigating Extreme-Token Phenomena in Transformers","created":"2025-10-10","cited_by_count":8,"area":null},{"id":"2601.06787","title":"Garbage Attention in Large Language Models: BOS Sink Heads and Sink-aware Pruning","created":"2026-01-11","cited_by_count":8,"area":null},{"id":"2601.21366","title":"Perceptrons and localization of attention's mean-field landscape","created":"2026-01-29","cited_by_count":8,"area":null},{"id":"2505.16284","title":"Only Large Weights (And Not Skip Connections) Can Prevent the Perils of Rank Collapse","created":"2025-05-22","cited_by_count":7,"area":null},{"id":"2507.02559","title":"Transformers Don't Need LayerNorm at Inference Time: Scaling LayerNorm Removal to GPT-2 XL and the Implications for Mechanistic Interpretability","created":"2025-07-03","cited_by_count":7,"area":"mechinterp"},{"id":"2510.10238","title":"The Achilles' Heel of LLMs: How Altering a Handful of Neurons Can Cripple Language Abilities","created":"2025-10-11","cited_by_count":7,"area":null},{"id":"2509.23676","title":"From Reasoning to Answer: Empirical, Attention-Based and Mechanistic Insights into Distilled DeepSeek R1 Models","created":"2025-09-28","cited_by_count":6,"area":"lr"},{"id":"2510.17021","title":"Forgetting to Forget: Attention Sink as A Gateway for Backdooring LLM Unlearning","created":"2025-10-19","cited_by_count":6,"area":"safety"},{"id":"2601.05866","title":"FACTUM: Mechanistic Detection of Citation Hallucination in Long-Form RAG","created":"2026-01-09","cited_by_count":6,"area":"rag"},{"id":"2505.16134","title":"Beyond Early-Token Bias: Model-Specific and Language-Specific Position Effects in Multilingual LLMs","created":"2025-05-22","cited_by_count":5,"area":null},{"id":"2602.01203","title":"Attention Sink Forges Native MoE in Attention Layers: Sink-Aware Training to Address Head Collapse","created":"2026-02-01","cited_by_count":5,"area":"moe"},{"id":"2508.03668","title":"CTR-Sink: Attention Sink for Language Models in Click-Through Rate Prediction","created":"2025-08-05","cited_by_count":4,"area":null},{"id":"2508.17807","title":"Attention Debiasing for Token Pruning in Vision Language Models","created":"2025-08-25","cited_by_count":4,"area":"mllm"},{"id":"2601.19657","title":"One Token Is Enough: Improving Diffusion Language Models with a Sink Token","created":"2026-01-27","cited_by_count":4,"area":"diffusion"},{"id":"2602.03510","title":"Semantic Routing: Exploring Multi-Layer LLM Feature Weighting for Diffusion Transformers","created":"2026-02-03","cited_by_count":4,"area":"diffusion"},{"id":"2602.05228","title":"Surgery: Mitigating Harmful Fine-Tuning for Large Language Models via Attention Sink","created":"2026-02-05","cited_by_count":4,"area":"alignment"},{"id":"2607.19139","title":"Text Template Tokens Are Implicit Semantic Registers in Diffusion Transformers","created":"2026-07-21","cited_by_count":4,"area":"diffusion"},{"id":"2507.04404","title":"LayerCake: Token-Aware Contrastive Decoding within Large Language Model Layers","created":"2025-07-06","cited_by_count":3,"area":null},{"id":"2508.16390","title":"A Large-Scale Benchmark for Evaluating Large Language Models on Medical Question Answering in Romanian","created":"2025-08-22","cited_by_count":3,"area":null},{"id":"2601.10159","title":"What Gets Activated: Uncovering Domain and Driver Experts in MoE Language Models","created":"2026-01-15","cited_by_count":3,"area":"moe"},{"id":"2603.06591","title":"What Makes Position Zero Special? A Mechanistic Study of Position Zero Attention Sinks in LLMs","created":"2026-02-04","cited_by_count":3,"area":null},{"id":"2603.15590","title":"Effective Distillation to Hybrid xLSTM Architectures","created":"2026-03-16","cited_by_count":3,"area":null},{"id":"2603.17771","title":"Attention Sinks Induce Gradient Sinks: Massive Activations as Gradient Regulators in Transformers","created":"2026-03-18","cited_by_count":3,"area":null},{"id":"2603.18017","title":"Frayed RoPE and Long Inputs: A Geometric Perspective","created":"2026-02-24","cited_by_count":3,"area":null},{"id":"2604.15701","title":"Improving Reasoning Capabilities in Small Models through Mixture-of-Layers Distillation with Stepwise Attention on Key Information","created":"2026-04-17","cited_by_count":3,"area":"lr"},{"id":"2605.09969","title":"The Truth Lies Somewhere in the Middle (of the Generated Tokens)","created":"2026-05-11","cited_by_count":3,"area":null},{"id":"2605.29657","title":"OccamToken: Efficient VLM Inference with Training-Free and Budget-Adaptive Token Pruning","created":"2026-05-28","cited_by_count":3,"area":"mllm"},{"id":"2507.09875","title":"Function Induction and Task Generalization: An Interpretability Study with Off-by-One Addition","created":"2025-07-14","cited_by_count":2,"area":null},{"id":"2509.22650","title":"RefAM: Attention Magnets for Zero-Shot Referral Segmentation","created":"2025-09-26","cited_by_count":2,"area":"diffusion"},{"id":"2510.19875","title":"Stream: Scaling up Mechanistic Interpretability to Long Context in LLMs via Sparse Attention","created":"2025-10-22","cited_by_count":2,"area":"mechinterp"},{"id":"2511.08028","title":"Generalizable Insights for Graph Transformers in Theory and Practice","created":"2025-11-11","cited_by_count":2,"area":null},{"id":"2512.01949","title":"Script: Graph-Structured and Query-Conditioned Semantic Token Pruning for Multimodal Large Language Models","created":"2025-12-01","cited_by_count":2,"area":"mllm"},{"id":"2602.17664","title":"Sink-Aware Pruning for Diffusion Language Models","created":"2026-02-19","cited_by_count":2,"area":"diffusion"},{"id":"2603.06248","title":"Gradient Flow Polarizes Softmax Outputs towards Low-Entropy Solutions","created":"2026-03-06","cited_by_count":2,"area":null},{"id":"2603.27898","title":"SAGE: Sink-Aware Grounded Decoding for Multimodal Hallucination Mitigation","created":"2026-03-29","cited_by_count":2,"area":"mllm"},{"id":"2604.10027","title":"SinkTrack: Attention Sink based Context Anchoring for Large Language Models","created":"2026-04-11","cited_by_count":2,"area":null},{"id":"2605.17887","title":"Attention Sinks and Outliers in Attention Residuals","created":"2026-05-18","cited_by_count":2,"area":null},{"id":"2606.02288","title":"Massive Spikes in LLMs are Bias Vectors: Mechanistic Uncovering and Spike-Free Quantization","created":"2026-06-01","cited_by_count":2,"area":null},{"id":"2608.28444","title":"Sliding-window beats linear attention","created":"2026-08-28","cited_by_count":2,"area":"inference"},{"id":"2505.19528","title":"AmpleHate: Amplifying the Attention for Versatile Implicit Hate Detection","created":"2025-05-26","cited_by_count":1,"area":null},{"id":"2510.08450","title":"gLSTM: Mitigating Over-Squashing by Increasing Storage Capacity","created":"2025-10-09","cited_by_count":1,"area":null},{"id":"2510.08870","title":"Quality Estimation Reranking for Document-Level Translation","created":"2025-10-10","cited_by_count":1,"area":null},{"id":"2510.16774","title":"Learning to play: A Multimodal Agent for 3D Game-Play","created":"2025-10-19","cited_by_count":1,"area":null},{"id":"2511.19895","title":"RPM-MCTS: Knowledge-Retrieval as Process Reward Model with Monte Carlo Tree Search for Code Generation","created":"2025-11-25","cited_by_count":1,"area":"lr"},{"id":"2512.09182","title":"Understanding the Failure Modes of Transformers through the Lens of Graph Neural Networks","created":"2025-12-09","cited_by_count":1,"area":null},{"id":"2602.00397","title":"Fast Forward: Accelerating LLM Prefill with Predictive FFN Sparsity","created":"2026-01-30","cited_by_count":1,"area":null},{"id":"2602.01984","title":"Enhancing Multi-Image Understanding through Delimiter Token Scaling","created":"2026-02-02","cited_by_count":1,"area":"mllm"},{"id":"2602.23179","title":"Induction Meets Biology: Mechanisms of Repeat Detection in Protein Language Models","created":"2026-02-26","cited_by_count":1,"area":null},{"id":"2604.10949","title":"Pseudo-Unification: Entropy Probing Reveals Divergent Information Patterns in Unified Multimodal Models","created":"2026-04-13","cited_by_count":1,"area":"lr"},{"id":"2605.06611","title":"The Structural Origin of Attention Sink: Variance Discrepancy, Super Neurons, and Dimension Disparity","created":"2026-05-07","cited_by_count":1,"area":null},{"id":"2605.08453","title":"Sink vs. diagonal patterns as mechanisms for attention switch and oversmoothing prevention","created":"2026-05-08","cited_by_count":1,"area":null},{"id":"2605.16331","title":"Retrieval and competition: how a protein foundation model starts a protein","created":"2026-05-05","cited_by_count":1,"area":null},{"id":"2606.20743","title":"Massive Activations Are Architecturally Robust: A Controlled Scratch/Commitment Residual Stream Test","created":"2026-06-17","cited_by_count":1,"area":null},{"id":"2607.13205","title":"Adaptive Filtering of the KV Cache: Diagnosing and Correcting Structural-Role Bias in LLM Inference","created":"2026-07-14","cited_by_count":1,"area":"inference"},{"id":"2608.11805","title":"Hybrid Gated Attention","created":"2026-08-12","cited_by_count":1,"area":null},{"id":"2608.11909","title":"Disentangling the Expressivity of RoPE","created":"2026-08-12","cited_by_count":1,"area":null},{"id":"2608.14712","title":"Which Question Is Your Attention Metric Answering? Attention Rows as Compositional Data","created":"2026-08-11","cited_by_count":1,"area":null},{"id":"2608.28383","title":"Semantic Head Specialization Guides Hybrid ViT Attention for Multimodal LLMs","created":"2026-08-28","cited_by_count":1,"area":"mllm"},{"id":"2609.00746","title":"Text Capability Loss in Vision-Language Adaptation: An Attention-Sink Diagnosis","created":"2026-09-01","cited_by_count":1,"area":"lr"},{"id":"2609.33659","title":"Learning Multimodal Embeddings with Evidence-Aligned Readout","created":"2026-09-27","cited_by_count":1,"area":"mllm"},{"id":"2505.07167","title":"Decoding One Safety Trigger Token for Balancing Safety and Usability in Large Language Models","created":"2025-05-12","cited_by_count":0,"area":"safety"},{"id":"2507.08214","title":"Depth-Sequence Transformer (DST) for Segment-Specific ICA Calcification Mapping on Non-Contrast CT","created":"2025-07-10","cited_by_count":0,"area":null},{"id":"2507.09071","title":"BlindSight: Harnessing Sparsity for Efficient Vision-Language Models","created":"2025-07-11","cited_by_count":0,"area":"mllm"},{"id":"2507.17618","title":"SimLens for Early Exit in Large Language Models: Eliciting Accurate Latent Predictions with One More Token","created":"2025-07-23","cited_by_count":0,"area":"inference"},{"id":"2508.17126","title":"Token Homogenization under Positional Bias","created":"2025-08-23","cited_by_count":0,"area":null},{"id":"2510.13860","title":"ShishuLM : Achieving Optimal and Efficient Parameterization with Low Attention Transformer Models","created":"2025-10-13","cited_by_count":0,"area":null},{"id":"2512.15600","title":"How Smoothing is N-simplicial Attention?","created":"2025-12-17","cited_by_count":0,"area":null},{"id":"2601.12430","title":"System-Mediated Attention Imbalances Make Vision-Language Models Say Yes","created":"2026-01-18","cited_by_count":0,"area":"mllm"},{"id":"2601.14522","title":"On the Runway Cascade of Transformers for Language Modeling","created":"2026-01-20","cited_by_count":0,"area":null},{"id":"2602.22419","title":"CLIP Is Shortsighted: Paying Attention Beyond the First Sentence","created":"2026-02-25","cited_by_count":0,"area":null},{"id":"2604.02973","title":"Exploring Motion-Language Alignment for Text-driven Motion Generation","created":"2026-04-03","cited_by_count":0,"area":null},{"id":"2604.10697","title":"Attention Sinks as Internal Signals for Hallucination Detection in Large Language Models","created":"2026-04-12","cited_by_count":0,"area":null},{"id":"2604.14722","title":"A Mechanistic Account of Attention Sinks in GPT-2: One Circuit, Broader Implications for Mitigation","created":"2026-04-16","cited_by_count":0,"area":null},{"id":"2605.09213","title":"Kinetic theory for Transformers and the lost-in-the-middle phenomenon","created":"2026-05-09","cited_by_count":0,"area":null},{"id":"2605.10503","title":"SLASH the Sink: Sharpening Structural Attention Inside LLMs","created":"2026-05-11","cited_by_count":0,"area":null},{"id":"2605.16468","title":"Mechanistically Interpretable Neural Encoding Reveals Fine-Grained Functional Selectivity in Human Visual Cortex","created":"2026-05-15","cited_by_count":0,"area":null},{"id":"2605.18599","title":"Resolving Representation Ambiguity in Feedforward Novel View Synthesis Transformer via Semantic-Spatial Decoupling","created":"2026-05-18","cited_by_count":0,"area":null}],"next_cursor":"100"}