PipeSD achieves 1.16x-2.16x speedup and 14.3%-25.3% lower energy use in cloud-edge LLM inference via token-batch pipeline scheduling optimized by dynamic programming and a Bayesian-optimized dual-threshold NAV trigger.
Speculative decoding: Exploiting speculative execution for accelerating seq2seq generation
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.DC 2years
2026 2verdicts
UNVERDICTED 2representative citing papers
SPECTRE achieves up to 2.28x speedup for large-model LLM serving by running speculative draft generation and target verification in parallel using idle tail-model services.
citing papers explorer
-
PipeSD: An Efficient Cloud-Edge Collaborative Pipeline Inference Framework with Speculative Decoding
PipeSD achieves 1.16x-2.16x speedup and 14.3%-25.3% lower energy use in cloud-edge LLM inference via token-batch pipeline scheduling optimized by dynamic programming and a Bayesian-optimized dual-threshold NAV trigger.
-
SPECTRE: Hybrid Ordinary-Parallel Speculative Serving for Resource-Efficient LLM Inference
SPECTRE achieves up to 2.28x speedup for large-model LLM serving by running speculative draft generation and target verification in parallel using idle tail-model services.