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
94 results for “keyword:"llm serving"” in ICLR 2027. 13 withdrawn or desk rejected listed apart. clearsort:relevanceratingnewestmost read
[PDF]Protect Before Evicting: Lagrangian-Guided Encoder Caching for Disaggregated Multimodal LLM Serving▼1361.33
[PDF]Analytical Provisioning for Attention–FFN Disaggregated LLM Serving under Stochastic Workloads▼4.3k2.00
[PDF]PluralGen: Efficient Test-Time Scaling by Decoupling Particle Search from Model Execution▼1.2k3.00
[PDF]StrataSpec: Self-Speculative Decoding with Structured Routes for Efficient LLM Serving▼5.8k2.33
[PDF]CandiKV: Preserving Candidate KV State Across Selection and Continuation in LLM Workflows▲3952.67
[PDF]SAS: Scaling Speculative Drafting for Efficient 1T LLM Serving with Semi-Autoregressive Decoding▲5103.00
[PDF]AgentReplay: Token-Wise Trace Replay Is Essential for Fair Serving System Performance Benchmarking▼1.2k3.00
[PDF]When the Second GPU Helps: Topology-Aware Mode Selection for Narrow-Interconnect LLM Serving▼1.3k1.75
[PDF]AdaptSpec: Input- and Step-Adaptive Tree-Based Speculative Decoding to Accelerate LLM Inference▼1.2k1.67
[PDF]LoRA-Ring: Locality-Preserving Consistent Hashing for Dynamic Adapter Routing in Large Language Models▲11.00
[PDF]Byte Economics: Million-Token Inference of a 2.8-Trillion-Parameter MoE on a Consumer Desktop▼12k1.67
[PDF]Speculating on the Past: Lossless Reuse of Stale KV Caches via Self-Speculative Verification▼4.3k2.00
Withdrawn or desk rejected (13)
[PDF]ReTree: Budget-Efficient Tree Speculative Decoding via Path Guidance and Sibling RecoveryWithdrawn1.67
[PDF]CACHECOHERENT: FREQUENCY-ALIGNED DOCUMENT ORDERING FOR PREFIX-CACHE-EFFICIENT MULTI-TENANT RAG SERVINGWithdrawn2.33
[PDF]SpecSys: Selective Speculative Decoding and Completion-Aware Scheduling for Efficient Agentic LLM InferenceWithdrawn2.00