UCSD Hao AI Lab 推出 JetSpec++，投机解码最高提速 7.45 倍
JetSpec++ 投机解码提速最高 7.45 倍
UCSD Hao AI Lab 推出投机解码方法 JetSpec++，通过因果并行草稿、隐状态预测器和锚点 token 解码器，在当前块验证前先草拟下一块，将草稿与验证流水线化。
Hao AI Lab· @haoailab · X·2026-10-31 02:46· 39 分钟前AI 评分42
AI 导读
UCSD Hao AI Lab 推出 JetSpec++，用因果并行草稿、隐状态预测器和锚点 token 解码器，在当前块验证前先草拟下一块，把草稿与验证流水线化，几乎隐藏全部草稿开销。
正文 · 原文
Introducing JetSpec++: we find causal parallel drafting paired with hidden states predictor and anchor token decoder can effectively draft the next future block before the current one is verified, and spending the freed time on self refinement.
This allows drafting to be pipelined with verification, hiding almost all of its drafting cost.
JetSpec++ reaches up to 1.54x end-to-end speedup over DSpark and 7.45x over autoregressive decoding while keeping lossless, at 1,361 TPS on Qwen3-8B and 367 TPS on Qwen3.8-27B in BF16.
With NVFP4 and kernel optimizations, JetSpec++ further reaches over 4,000 TPS on Qwen3-8B and over 900 TPS on Qwen3.8-27B, on two GB200 GPUs. ⚡️
Project page: https://jetspec-project.github.io/jetspecpp-web/
来源：Hao AI Lab · x.com
来源：RadarAI · aihot.news