deepseek-v4.1THREAD
V4.1-Flash-1017 本地 3bit 量化跑到 100 tok/s，作者发布首个 HF 仓库
作者为 V4.1-Flash-1017 做了 3bit 量化包，本地推理约 100 tok/s，并称这是自己第一个 Hugging Face 仓库。
Nate’s in SF
@Nateemerson12 天前收录原推发布于 2026.10.18
THREAD
I built a pack for DeepSeek-v4.1-Flash-1017 Running 3bit at 100 tok/s locally My first huggingface repo!
这条是引用着另一条写的：x.com/deepseek_ai/status/2111375640717033777
Working on some AB tests with real benches now. Please let me know if you see anything I did wrong or should fix, hoping to do more of this in the future! https://huggingface.co/Nateemerson/DeepSeek-V4.1-Flash-1017-EXL3-3bpw-2x-RTX-PRO-6000
Nate’s in SF @Nateemerson原推发布于 2026.10.18
x.com/nateemerson/status/2111693164704944222采集于 2026.10.18 16:24