{"total":74,"rows":[{"id":"2510.00636","title":"Expected Attention: KV Cache Compression by Estimating Attention from Future Queries Distribution","created":"2025-10-01","cited_by_count":74,"area":"inference"},{"id":"2509.26488","title":"dParallel: Learnable Parallel Decoding for dLLMs","created":"2025-09-30","cited_by_count":65,"area":"inference"},{"id":"2601.07568","title":"d3LLM: Ultra-Fast Diffusion LLM using Pseudo-Trajectory Distillation","created":"2026-01-12","cited_by_count":37,"area":"inference"},{"id":"2504.06261","title":"Hogwild! Inference: Parallel LLM Generation via Concurrent Attention","created":"2025-04-08","cited_by_count":32,"area":"inference"},{"id":"2504.18583","title":"PARD: Accelerating LLM Inference with Low-Cost PARallel Draft Model Adaptation","created":"2025-04-23","cited_by_count":29,"area":"inference"},{"id":"2604.12989","title":"Accelerating Speculative Decoding with Block Diffusion Draft Trees","created":"2026-04-14","cited_by_count":29,"area":"inference"},{"id":"2509.23094","title":"d$^2$Cache: Accelerating Diffusion-Based LLMs via Dual Adaptive Caching","created":"2025-09-27","cited_by_count":25,"area":"inference"},{"id":"2601.19278","title":"DART: Diffusion-Inspired Speculative Decoding for Fast LLM Inference","created":"2026-01-27","cited_by_count":24,"area":"inference"},{"id":"2603.04445","title":"Dynamic Model Routing and Cascading for Efficient LLM Inference: A Survey","created":"2026-02-23","cited_by_count":19,"area":"inference"},{"id":"2412.17483","title":"A Silver Bullet or a Compromise for Full Attention? A Comprehensive Study of Gist Token-based Context Compression","created":"2024-12-23","cited_by_count":17,"area":"inference"},{"id":"2605.18810","title":"D-PACE: Dynamic Position-Aware Cross-Entropy for Parallel Speculative Drafting","created":"2026-05-12","cited_by_count":17,"area":"inference"},{"id":"2605.07721","title":"Memory-Efficient Looped Transformer: Decoupling Compute from Memory in Looped Language Models","created":"2026-05-08","cited_by_count":14,"area":"inference"},{"id":"2509.18362","title":"FastMTP: Accelerating LLM Inference with Enhanced Multi-Token Prediction","created":"2025-09-16","cited_by_count":13,"area":"inference"},{"id":"2605.29343","title":"Draft-OPD: On-Policy Distillation for Speculative Draft Models","created":"2026-05-28","cited_by_count":11,"area":"inference"},{"id":"2502.00085","title":"Efficient Beam Search for Large Language Models Using Trie-Based Decoding","created":"2025-01-31","cited_by_count":10,"area":"inference"},{"id":"2605.08632","title":"PARD-2: Target-Aligned Parallel Draft Model for Dual-Mode Speculative Decoding","created":"2026-05-09","cited_by_count":8,"area":"inference"},{"id":"2503.24358","title":"SQuat: Subspace-orthogonal KV Cache Quantization","created":"2025-03-31","cited_by_count":7,"area":"inference"},{"id":"2505.15774","title":"Beyond Hard and Soft: Hybrid Context Compression for Balancing Local and Global Information Retention","created":"2025-05-21","cited_by_count":7,"area":"inference"},{"id":"2508.15212","title":"SparK: Query-Aware Unstructured Sparsity with Recoverable KV Cache Channel Pruning","created":"2025-08-21","cited_by_count":6,"area":"inference"},{"id":"2606.01813","title":"Cost-Aware Diffusion Draft Trees for Speculative Decoding","created":"2026-06-01","cited_by_count":6,"area":"inference"},{"id":"2608.13524","title":"DARTree: Speculative Diffusion Decoding with Autoregressive Draft Trees","created":"2026-08-13","cited_by_count":6,"area":"inference"},{"id":"2511.11907","title":"KVSwap: Disk-aware KV Cache Offloading for Long-Context On-device Inference","created":"2025-11-14","cited_by_count":5,"area":"inference"},{"id":"2608.30135","title":"Verification-Aware Training for Speculative Decoding","created":"2026-08-31","cited_by_count":5,"area":"inference"},{"id":"2509.10798","title":"Judge Q: Trainable Queries for Optimized Information Retention in KV Cache Eviction","created":"2025-09-13","cited_by_count":4,"area":"inference"},{"id":"2510.13602","title":"NOSA: Native and Offloadable Sparse Attention","created":"2025-10-15","cited_by_count":4,"area":"inference"},{"id":"2606.03819","title":"TreeFlash: Parallel AR-Approximation for Faster Speculative Decoding","created":"2026-06-02","cited_by_count":4,"area":"inference"},{"id":"2607.19223","title":"AdaFlash: Adaptive Speculative Decoding via On-Policy Distilled Diffusion Drafters","created":"2026-07-21","cited_by_count":4,"area":"inference"},{"id":"2605.09329","title":"Test-Time Speculation","created":"2026-05-10","cited_by_count":3,"area":"inference"},{"id":"2607.06763","title":"Trees from Marginals: Autoregressive drafting with factorized priors","created":"2026-07-07","cited_by_count":3,"area":"inference"},{"id":"2608.02123","title":"From Chains to Trees: Parent-Conditioned Drafting for Semi-Autoregressive Speculative Decoding","created":"2026-08-03","cited_by_count":3,"area":"inference"},{"id":"2504.09775","title":"MIST: A Co-Design Framework for Heterogeneous, Multi-Stage LLM Inference","created":"2025-04-14","cited_by_count":2,"area":"inference"},{"id":"2605.24793","title":"Beyond the Target: From Imitation to Collaboration in Speculative Decoding","created":"2026-05-24","cited_by_count":2,"area":"inference"},{"id":"2606.31315","title":"BlockPilot: Instance-Adaptive Policy Learning for Diffusion-based Speculative Decoding","created":"2026-06-30","cited_by_count":2,"area":"inference"},{"id":"2608.04074","title":"Spend Bits Where Queries Look: KV Cache Vector Quantization with Attention-Preserving Transforms","created":"2026-08-04","cited_by_count":2,"area":"inference"},{"id":"2609.17184","title":"LoopSpec: Pipelined Self-Speculative Decoding for Looped Transformers","created":"2026-09-15","cited_by_count":2,"area":"inference"},{"id":"2609.27396","title":"When Parallel Drafter Meets Parallel Speculative Decoding","created":"2026-09-23","cited_by_count":2,"area":"inference"},{"id":"2609.29812","title":"FlashLoop: Fast and Memory-Efficient Looped Transformers via Lazy Updates","created":"2026-09-24","cited_by_count":2,"area":"inference"},{"id":"2601.23180","title":"TriSpec: Ternary Speculative Decoding via Lightweight Proxy Verification","created":"2026-01-30","cited_by_count":1,"area":"inference"},{"id":"2603.05210","title":"Balancing Coverage and Draft Latency in Vocabulary Trimming for Faster Speculative Decoding","created":"2026-03-05","cited_by_count":1,"area":"inference"},{"id":"2605.01910","title":"Stochastic Sparse Attention for Memory-Bound Inference","created":"2026-05-03","cited_by_count":1,"area":"inference"},{"id":"2605.18165","title":"Elastic-dLLM: Position Preserving Context Compression and Augmentation of Diffusion LLMs","created":"2026-05-18","cited_by_count":1,"area":"inference"},{"id":"2606.05742","title":"AdaPLD: Adaptive Retrieval and Reuse for Efficient Model-Free Speculative Decoding","created":"2026-06-04","cited_by_count":1,"area":"inference"},{"id":"2608.11742","title":"Ripple-Pivot Search: Active Parallel Decoding for Diffusion Large Language Models","created":"2026-08-12","cited_by_count":1,"area":"inference"},{"id":"2608.13925","title":"CForce: Boosting Parallel Decoding for dLLMs via Consistency Forcing","created":"2026-08-14","cited_by_count":1,"area":"inference"},{"id":"2608.14191","title":"KV Cache Compression Through the Lens of Transform Coding","created":"2026-08-14","cited_by_count":1,"area":"inference"},{"id":"2609.02897","title":"Margins, Not Windows: Training-Free Per-Step Lossy Speculative Decoding","created":"2026-07-03","cited_by_count":1,"area":"inference"},{"id":"2610.06479","title":"Behavior-Preserving KV Cache Compression","created":"2026-10-05","cited_by_count":1,"area":"inference"},{"id":"2602.16994","title":"Dynamic Delayed Tree Expansion For Improved Multi-Path Speculative Decoding","created":"2026-02-19","cited_by_count":0,"area":"inference"},{"id":"2605.08151","title":"SPECTRE: Hybrid Ordinary-Parallel Speculative Serving for Resource-Efficient LLM Inference","created":"2026-05-04","cited_by_count":0,"area":"inference"},{"id":"2605.26558","title":"Cassandra: Enabling Reasoning LLMs at Edge via Self-Speculative Decoding","created":"2026-05-26","cited_by_count":0,"area":"inference"},{"id":"2608.08721","title":"LibraSpec: Dynamic Diffusion-Based Speculative Decoding via Marginal-Gain-Driven Optimization","created":"2026-08-09","cited_by_count":0,"area":"inference"},{"id":"2608.14787","title":"From Positionwise Confidence to Prefix Scheduling: Verifier Skipping in Speculative Decoding","created":"2026-08-14","cited_by_count":0,"area":"inference"},{"id":"2609.16450","title":"Early-Bird Decoding: Accelerating Diffusion LLMs with Learnable Block Sizes and Parallel Sampling","created":"2026-09-15","cited_by_count":0,"area":"inference"},{"id":"2609.23033","title":"WaveFront Decoding: Parallelized Self-Speculative Decoding for Looped Language Models","created":"2026-09-19","cited_by_count":0,"area":"inference"},{"id":"2609.23314","title":"ValueDiff: Value-Geometric KV Cache Eviction for Sink-Suppressed LLMs","created":"2026-09-20","cited_by_count":0,"area":"inference"},{"id":"2609.33652","title":"Tsubame: Tree Replay for Diffusion-Based Speculative Decoding","created":"2026-09-27","cited_by_count":0,"area":"inference"},{"id":"2609.34388","title":"Reciprocal Guidance: Orchestrating Draft and Verify Budgets for Advancing the Diffusion-AR Self-Speculation Frontier","created":"2026-09-28","cited_by_count":0,"area":"inference"},{"id":"2609.34832","title":"BV Loss: Block Verification-Aware Loss for Block Diffusion Speculative Decoding","created":"2026-09-28","cited_by_count":0,"area":"inference"},{"id":"2609.37029","title":"LongSpark: Efficient speculative decoding with a fixed-cost parallel drafter","created":"2026-09-29","cited_by_count":0,"area":"inference"},{"id":"2609.37532","title":"DScale: Scaling Block-Diffusion Speculative Decoding with Adaptive Verification","created":"2026-09-29","cited_by_count":0,"area":"inference"},{"id":"2609.38121","title":"WUSH-KV: KV Cache Quantization with Data-Adaptive Transforms","created":"2026-09-29","cited_by_count":0,"area":"inference"},{"id":"2609.38169","title":"STEPQuant: When and Where Errors Matter in Delta-Rule Recurrent State Quantization","created":"2026-09-29","cited_by_count":0,"area":"inference"},{"id":"2609.38510","title":"DEdit: Iterative Draft Editing for Speculative Decoding","created":"2026-09-29","cited_by_count":0,"area":"inference"},{"id":"2609.39972","title":"UBTree: Parallel Tree Drafting via Unigram and Bigram Models for Speculative Decoding","created":"2026-09-30","cited_by_count":0,"area":"inference"},{"id":"2610.00321","title":"CAST: Cost-Aware Speculative Trees from One-Pass Block Drafters","created":"2026-09-29","cited_by_count":0,"area":"inference"},{"id":"2610.01439","title":"DRelay: Global Draft Context for Prefix-Aware Parallel Speculative Decoding Repair","created":"2026-10-01","cited_by_count":0,"area":"inference"},{"id":"2610.02713","title":"WakeKV: Reactive, Reversible KV Residency for Heads That Change Their Minds","created":"2026-10-02","cited_by_count":0,"area":"inference"},{"id":"2610.03027","title":"Tailoring the Quantization Space for 1-Bit KV Cache Compression","created":"2026-10-02","cited_by_count":0,"area":"inference"},{"id":"2610.05106","title":"SharpDraft: Accelerating Long-Context Speculative Decoding with Cardinality-Aware Query Scaling","created":"2026-10-04","cited_by_count":0,"area":"inference"},{"id":"2610.07659","title":"DLoop: Looped Speculative Decoding","created":"2026-10-06","cited_by_count":0,"area":"inference"},{"id":"2610.12570","title":"Contextual Correction of Parallel Draftsfor Speculative Decoding","created":"2026-10-08","cited_by_count":0,"area":"inference"},{"id":"2610.14448","title":"KV4-RL: Unlocking Faster Asynchronous Rollouts for RLVR with INT4 KV Cache Quantization","created":"2026-10-10","cited_by_count":0,"area":"inference"},{"id":"2610.19026","title":"Joint Subspace Sparse Attention for Efficient Long-Context Decoding","created":"2026-10-13","cited_by_count":0,"area":"inference"},{"id":"2610.20175","title":"TreeSpark: Learning to Draft the Trees You Verify for Speculative Decoding","created":"2026-10-14","cited_by_count":0,"area":"inference"}],"next_cursor":null}