StillDeadcode/radiance-qwen-image-2.1-iq8r

Qwen-Image 2.1, IQ8R — radiance container

Qwen/Qwen-Image-2.1 — text to image and image editing — as a single .rad container for the radiance inference engine (AMD RDNA4, ROCm): the Qwen3-VL-8B conditioning network with its vision tower, the 7.1B diffusion transformer and the RGBA VAE.

It replaces StillDeadcode/radiance-qwen-image-2.1-fp8: the same eight bits a weight, closer to the bf16 model's pictures.

File	qwen-image-2.1-iq8r.rad — 16.13 GiB
Weights	the conditioning network's (Qwen3-VL-8B) and the diffusion transformer's linears int8 (codes −127..127) with a u8 scale per row and 256 of K under a bf16 scale a row, each 128 of K rotated by the unnormalised Hadamard, round to nearest; the vision tower, the small linears, the norms and the VAE bf16
Activations	int8 under the same rotation (W8A8)
Steps	40 by default (num_inference_steps, or quality: low / medium / high)
Needs	radiance built from main after 1.17.0: IQ8R at one scale a 256 of K is not in 1.17.0 and comes with the next release

Serve

# radiance.yaml
server:
  host: 0.0.0.0
  port: 8000
models:
  qwen-image:
    model: qwen-image-2.1-iq8r.rad
    tp: 2

radiance --config radiance.yaml

tp: 1 serves it on one card. The server speaks OpenAI's Images API: POST /v1/images/generations and POST /v1/images/edits (JSON, or multipart with image / image[] and mask, as OpenAI's SDKs send them). Pictures come back as b64_json (png, jpeg or webp; response_format: url is refused). Also: n, size (WxH or auto = 2048x2048), seed, streaming with partial_images, background: transparent; edits take 1 to 10 reference pictures and a mask. /v1/models lists the model.

Seconds a 1024x1024 picture at 20 steps on Radeon AI PRO R9700 (gfx1201): 10.5 on one card and 6.8 on two (the FP8 container: 10.2 and 6.6).

Against radiance's bf16 container of the same model from the same noise (four 1024x1024 pictures, 20 steps), PSNR 32.6 dB on one card and 34.1 dB on two; the FP8 container is at 28.1 and 28.6 dB.

How this file was made

rad-convert Qwen/Qwen-Image-2.1 --recipe qwen-image-2.1-iq8r.recipe -o qwen-image-2.1-iq8r.rad

Qwen/Qwen-Image-2.1 is the diffusers download of that repository.

The recipe (qwen-image-2.1-iq8r.recipe in this repository) names what is quantised; everything it does not name is the checkpoint's own dtype:

te.blk.*.attn_qkv.weight             rtn  codes=i8 group=256 scale=u8 scale2=bf16 block2=1x* transform=fwht128 clamp=sym rule=search
te.blk.*.attn_output.weight          rtn  codes=i8 group=256 scale=u8 scale2=bf16 block2=1x* transform=fwht128 clamp=sym rule=search
te.blk.*.ffn_gate_up.weight          rtn  codes=i8 group=256 scale=u8 scale2=bf16 block2=1x* transform=fwht128 clamp=sym rule=search
te.blk.*.ffn_down.weight             rtn  codes=i8 group=256 scale=u8 scale2=bf16 block2=1x* transform=fwht128 clamp=sym rule=search
dit.blk.*.attn_qkv.weight            rtn  codes=i8 group=256 scale=u8 scale2=bf16 block2=1x* transform=fwht128 clamp=sym rule=search
dit.blk.*.attn_output.weight         rtn  codes=i8 group=256 scale=u8 scale2=bf16 block2=1x* transform=fwht128 clamp=sym rule=search
dit.blk.*.ffn_gate_up.weight         rtn  codes=i8 group=256 scale=u8 scale2=bf16 block2=1x* transform=fwht128 clamp=sym rule=search
dit.blk.*.ffn_down.weight            rtn  codes=i8 group=256 scale=u8 scale2=bf16 block2=1x* transform=fwht128 clamp=sym rule=search
vae.*                                cast dtype=bf16

License

Built with Qwen. Qwen-Image 2.1 is licensed under the Qwen RESEARCH LICENSE AGREEMENT (non-commercial; LICENSE in this repository), and NOTICE carries its attribution. This container is a modified copy: its linear weights are re-encoded as rotated int8 (see above).