DeepSeek Releases DSpark, a Speculative Decoding Framework That Accelerates DeepSeek-V4 Per-User Generation 60–85% Over MTP-1

DeepSeek open-sourced DSpark, a speculative decoding framework that accelerates per-user generation for DeepSeek-V4 by 57–85% over the MTP-1 baseline. The framework pairs a parallel draft backbone with a Markov head and confidence-scheduled verification to reduce suffix decay and adapt to real-time GPU load. DSpark achieves lossless speedups with accepted length gains of 16–31% over prior methods.

DeepSeek open-sourced DSpark, a speculative decoding framework that attaches a draft module to existing DeepSeek-V4 weights. It pairs a parallel draft backbone with a lightweight Markov head to cut suffix decay, then adds confidence-scheduled verification that tailors how many tokens get checked to real-time GPU load. Offline, accepted length rises 16–31% over DFlash and Eagle3; in production it speeds per-user generation 57–85% over the MTP-1 baseline, losslessly. The training repo, DeepSpec, ships under MIT. The post DeepSeek Releases DSpark, a Speculative Decoding Framework That Accelerates DeepSeek-V4 Per-User Generation 60–85% Over MTP-1 https://www.marktechpost.com/2026/06/27/deepseek-releases-dspark-a-speculative-decoding-framework-that-accelerates-deepseek-v4-per-user-generation-60-85-over-mtp-1/ appeared first on MarkTechPost https://www.marktechpost.com .