NVIDIA releases 4B speculative decoding draft boosting Qwen3.8-27B acceptance by 46%
_akhaliq · x · 2026-11-18
NVIDIA has published a speculative decoding draft model on Hugging Face: a 4B-parameter draft designed to accelerate Qwen3.8-27B, improving token acceptance by 46% over the model's built-in drafting head. It's an official option for speeding up inference in self-hosted deployments.
More from Infra
- Perplexity launches Orchestrator API: 45+ models, auto-routing, unified billing — perplexity_ai · 2026-11-18
- Replit's agent swarm rebuilds an $83,700/yr observability tool as open-source DataCat — amasad · 2026-11-18
- Google Cloud switches to threshold billing, charging mid-month as costs hit thresholds — CtrlAltDwayne · 2026-11-18
- Google placed furthest on vision in Gartner's 2026 Cloud DB Magic Quadrant — rseroter · 2026-11-18
- Cerebras CEO: 75% of data centers will be outside the US by 2027 after midterm results — johncoogan · 2026-11-18
- CoreWeave faces $19.05B principal-plus-interest bill by end of 2027 — SumitGup · 2026-11-18