SmartApe AI
Nov 22, 2026 · Artificial Intelligence
Tencent's DSpine Accelerates Qwen3.8-27B 3.84x with 2B Draft Model
Tencent open-sources DSpine, a speculative decoding framework that uses a 2B draft model to accelerate Qwen3.8-27B, achieving 3.84x throughput on WorkBuddy Office via adjacent causal injection and transition cache, outperforming DFlash2 across math, coding, and agent benchmarks.
Adjacent Causal InjectionBenchmarkDSpine
0 likes · 7 min read
