qwen3.5-2b for the Ryzen AI NPU

Qwen3.5-2B, converted to run on the Ryzen AI NPU of AMD Strix Halo inside halogen-flash-server, beside the Flash model on the GPU and behind the same port. Questions go to the Discord. Bugs go to the server repository's issues.

These files run only inside the halogen-flash-server container, on the NPU. They will not load in transformers or any other runtime.

What is here

devices/              75 MB  the NPU program for this model, and its NOTICE.md
qwen3.5-2b.hnpw    3.6 GB  the weights, converted for the NPU (4-bit for generation)
tokenizer/            13 MB  the tokenizer

Use it

Start halogen-flash-server 0.16.2 or later with HALOGEN_DOWNLOAD on and HALOGEN_NPU_MODELS=qwen3.5-2b. The server fetches these files into /models/npu/qwen3.5-2b/ and checks each one's size and checksum before it starts. It answers /v1/chat/completions for requests that name qwen3.5-2b, thinking off, text only: summaries and other short jobs beside the Flash model. A prompt is up to 16,384 tokens.

The files at the head of this repository need halogen-flash-server 0.18.12 or later; each server version downloads the revision it was released with, so an older server keeps working unchanged.

The host needs the NPU driver, XRT with its NPU plugin, and the GPU's fabric clock held while the NPU works beside the GPU. The setup, every request shape, are in docs/NPU.md.

Source and license

Converted from Qwen/Qwen3.5-2B at revision 15852e8c16360a2fea060d615a32b45270f8a8fc. Apache-2.0, as the source. devices/NOTICE.md lists the third-party notices the NPU program carries.