Recursive Speculative Decoding: Accelerating LLM Inference via Sampling Without Replacement

arXiv:2402.14160v1 Announce Type: new
Abstract: Speculative decoding is an inference-acceleration method for large language models (LLMs) where a small language model generates a draft-token sequence which is further verified by the target LLM in parallel. Recent works have advanced this method by establishing a draft-token tree, achieving superior performance over a single-sequence speculative decoding. However, those works independently generate tokens at each level of the tree, not leveraging the tree's entire diversifiability. Besides, their empirical superiority has been shown for …

abstract advanced arxiv cs.ai cs.lg decoding draft inference language language model language models large language large language models llm llms performance recursive replacement sampling small small language model token tree type via