nerv-dynamics/Qwen3.8-27B-S2

Qwen3.8-27B for S2

Qwen/Qwen3.8-27B quantised to NVFP4 for S2, the inference engine of Nerv Linux, with the Inco DFlash2 drafter for speculative decoding.

file	what it is
Qwen3.8-27B-s2v9dp.s2w	the model: NVFP4 weights in S2's format, with its tokenizer and chat format
Qwen3.8-27B-dflash2-inco.s2w	the drafter, in S2's format

These files are S2's own format (.s2w), not safetensors: they run on S2 and nothing else.

Run it

On Nerv:

dnf install nerv-s2
s2 pull qwen3.8-27b      # both files, checked against the SHA-256 the package records
s2                       # the dashboard: n, then Enter

The first run prepares the model for the machine's GPU, once. The dashboard serves OpenAI's chat completions API at http://127.0.0.1:8000/v1 unless told not to.

Licence

Apache-2.0, as Qwen3.8-27B and the Inco drafter are. Both are converted: the weights are quantised from Qwen's release, and the drafter is Inco's, repacked. LICENSE is Qwen's licence text.