deepseek-v4.1POST

Unsloth 发布 V4.1-Flash-1017 的 1-bit GGUF：233–250GB

1-bit 量化体积 233–250GB（Q8 为 508GB），85–88% 情况下与全模型首选 token 一致；233GB 版可在 128GB Mac 上跑未合并的 llama.cpp PR，111GB n-gram 表放 SSD。

kaimi

@kaimi_local11 天前收录原推发布于 2026.10.20

kaimi @kaimi_local原推发布于 2026.10.20

Unsloth put out 1-bit GGUFs of DeepSeek V4.1-Flash-1017 at 233-250GB, vs 508GB at Q8, and they pick the full model's top token 85-88% of the time. The 233GB one runs on a 128GB Mac with an unmerged llama.cpp PR, which keeps its 111GB n-gram table on SSD.

这条是引用着另一条写的：x.com/unslothai/status/2112212605930778830

x.com/kaimi_local/status/2112367688169173181采集于 2026.10.20 12:02