An Efficient Inference Framework for Early-exit Large Language Models
arxiv(2024)
Abstract
Building efficient inference framework has gained increasing interests for
research community. Early-exit models, a variant of LLMs, improves the
inference efficiency of LLMs by skipping rest layers and directly generate
output tokens when they are confident enough. However, there is no work of LLM
inference framework that takes early-exit models into consideration. This is
non-trivial as prior art on LLM inference cannot be directly applied to
early-exit models. In this work, we solves two key challenges in building
efficient inference framework for early-exit models: (1) batch inference at
iteration-level granularity; and (2) KV cache management. For the former, we
propose to process the batch until all sequences surpass the early-exit
confidence threshold. For the latter, we propose to fill the KV cache of rest
layers before the iteration terminates. Our evaluation shows that, compared
with the original vLLM operating at full layers, our solution achieves up to
1.25x speed up.
MoreTranslated text
AI Read Science
Must-Reading Tree
Example
![](https://originalfileserver.aminer.cn/sys/aminer/pubs/mrt_preview.jpeg)
Generate MRT to find the research sequence of this paper
Chat Paper
Summary is being generated by the instructions you defined