Qwen3.8-27B is now up to ~3× faster on Apple Silicon with mlx-dspark
英文标题：Qwen3.8-27B is now up to ~3× faster on Apple Silicon with mlx-dspark
mlx-dspark is an MLX port of DeepSeek's DSpark speculative-decoding drafters (the DeepSpec release), plus z-lab's DFlash, with one lossless verify loop. v0.10.0 adds Qwen3.8-27B via RadixArk's drafter, the first SpecForge/SGLang-packaged head it loads. Numbers (M4 Pro 48 GB, medians of 3, gree
打开原文相关 2 · 机会 0
中文
mlx-dspark is an MLX port of DeepSeek's DSpark speculative-decoding drafters (the DeepSpec release), plus z-lab's DFlash, with one lossless verify loop. v0.10.0 adds Qwen3.8-27B via RadixArk's drafter, the first SpecForge/SGLang-packaged head it loads. Numbers (M4 Pro 48 GB, medians of 3, gree
中文全文翻译正在补齐，可先查看下方英文原文。
English / 原文
mlx-dspark is an MLX port of DeepSeek's DSpark speculative-decoding drafters (the DeepSpec release), plus z-lab's DFlash, with one lossless verify loop. v0.10.0 adds Qwen3.8-27B via RadixArk's drafter, the first SpecForge/SGLang-packaged head it loads. Numbers (M4 Pro 48 GB, medians of 3, gree