NVIDIA Releases Draft Model to Accelerate Speculative Decoding
NVIDIA released a 4B-parameter Qwen3.5-4B draft model on Hugging Face for Qwen3.8-27B speculative decoding, boosting token acceptance by 46% over the model's built-in drafting.
2026-11-18 ~ 2026-11-19 · 2 related posts
- NVIDIA releases 4B speculative decoding draft boosting Qwen3.8-27B acceptance by 46% — _akhaliq · 2026-11-18
- NVIDIA's Qwen3.5-4B draft model for Qwen3.8-27B trends on Hugging Face — nvidia · 2026-11-19