Speculative Prefill: Turbocharging TTFT with Lightweight and Training-Free Token Importance Estimation

Jingyu Liu; Beidi Chen; Ce Zhang

2025 ICML ICML 2025

Speculative Prefill: Turbocharging TTFT with Lightweight and Training-Free Token Importance Estimation

Abstract

No abstract available. No abstract was captured for this paper when its proceedings were scraped - most often an older edition published without one online. How the corpus is built.

Authors

Jingyu Liu, Beidi Chen, Ce Zhang

Topics

Machine Learning > Application Areas > Efficient Computing Artificial Intelligence > Core AI > Large Language Models

Keywords

lightweight model llm inference fp8 speculative prefill token importance estimation ttft

Download PDF