NeurIPS 2026 reviews are up: reviews, scores, rebuttals and meta-reviews for 7,877 papers. See the highest rated
acceptodds: the papers of ICLR 2027
106 results for “keyword:"inference acceleration"” in ICLR 2027. 12 withdrawn or desk rejected listed apart. clearsort:relevanceratingnewestmost read
[PDF]Frequencies Rank, Probabilities Allocate: Calibrated Draft Trees for Drafter-Free Speculative Decoding▲3972.67
[PDF]LatentSpec: Two-Stage Thinker–Drafter Parallel Drafting with Shared Latent Reasoning for Speculative Decoding▲9.2k2.67
[PDF]When to Trust the Past: Reliability-Calibrated Draft-Tree Orchestration for Speculative Decoding▲3972.67
[PDF]SharpDraft: Accelerating Long-Context Speculative Decoding with Cardinality-Aware Query Scaling▲3.0k2.33
[PDF]Truncate to Accelerate: From Forecast Recoverability to Efficient Time-Series Foundation Models▼4.5k2.67
[PDF]Accelerating Decoding via Autoregressive-to-Parallel Distillation for Architecture-Agnostic Generative Recommendation▼4.3k2.00
[PDF]Fork and Regenerate: MTP-Guided Early Forking and Suffix Regeneration for Block Diffusion Speculative Decoding▲3.0k2.33
[PDF]Phase-MoE: A Co-Design to Bound the Expert Explosion in Block Diffusion Language Models▼4.3k2.00
[PDF]RoboWorM: Evaluating Movement Realism, Consistency and Acceleration Affects in Embodied Video World Models▲3972.67
[PDF]ReSPect: Residual Spectral Prediction for Training-Free Acceleration of Diffusion Transformers▲3.0k2.33
[PDF]The Shape of Speculation: Efficient Draft Topology for Speculative Jacobi Decoding in Autoregressive Image Generation▼5.8k2.33
[PDF]One Patch, Three Roles: What Is Actually Coupled in Autoregressive Time-Series Forecasting?▼5.9k2.50
[PDF]Correct While Verifying: Intermediate Target Predictions for Speculative Draft Correction▼1361.33
[PDF]ASSIST-VLA: ACCELERATING VISION-LANGUAGE- ACTION INFERENCE VIA MULTI-LEVEL REDUN- DANCY REDUCTION▲6.1k2.00
[PDF]One Objective, One Stage: Unifying Relevance, Coverage, and Redundancy in Visual Token Pruning▼4.3k2.00
[PDF]MAX-Cache: Motion-Aligned Activation Reuse with Mismatch-Guided Selective Recomputation for Efficient Autoregressive Video Diffusion▲3972.67
[PDF]ActionCache: Training-Free Acceleration for Vision-Language-Action Models with Action Caching and Refinement▼4.5k2.25