From 636bd2e4710cdfc6f77c5bde9deb6259b5559d00 Mon Sep 17 00:00:00 2001 From: Vladimir Mandic Date: Wed, 27 Aug 2025 12:23:33 -0400 Subject: [PATCH] add Nunchaku-Qwen-Image-Lightning Signed-off-by: Vladimir Mandic --- CHANGELOG.md | 3 +++ html/reference.json | 2 +- pipelines/qwen/qwen_nunchaku.py | 4 +++- 3 files changed, 7 insertions(+), 2 deletions(-) diff --git a/CHANGELOG.md b/CHANGELOG.md index 9b16805d4..e5d5833ef 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -6,6 +6,9 @@ - **Chroma** final versions: [Chroma1-HD](https://huggingface.co/lodestones/Chroma1-HD), [Chroma1-Base](https://huggingface.co/lodestones/Chroma1-Base) and [Chroma1-Flash](https://huggingface.co/lodestones/Chroma1-Flash) - **Qwen-Image** [InstantX ControlNet Union](https://huggingface.co/InstantX/Qwen-Image-ControlNet-Union) support *note* qwen-image is already a very large model and controlnet adds 3.5GB on top of that so quantization and offloading are highly recommended! + - [Nunchaku-Qwen-Image-Lightning](https://huggingface.co/nunchaku-tech/nunchaku-qwen-image) + if you have a compatible nVidia GPU, Nunchaku is the fastest quantization engine, currently available for Flux.1, SANA and Qwen-Image models + *note*: release version of `nunchaku==0.3.2` does NOT include support, so you need to build [nunchaku](https://nunchaku.tech/docs/nunchaku/installation/installation.html) from source - updated [SD.Next Model Samples Gallery](https://vladmandic.github.io/sd-samples/compare.html) - **Core** - enable offload during pre-forward by default diff --git a/html/reference.json b/html/reference.json index a7f5e214d..2425350f1 100644 --- a/html/reference.json +++ b/html/reference.json @@ -196,7 +196,7 @@ "skip": true, "extras": "" }, - "Qwen-Lightning": { + "Qwen-Image-Lightning": { "path": "vladmandic/Qwen-Lightning", "preview": "Qwen-Lightning.jpg", "desc": " Qwen-Lightning is step-distilled from Qwen-Image to allow for generation in 8 steps.", diff --git a/pipelines/qwen/qwen_nunchaku.py b/pipelines/qwen/qwen_nunchaku.py index 698b59447..4464a12b4 100644 --- a/pipelines/qwen/qwen_nunchaku.py +++ b/pipelines/qwen/qwen_nunchaku.py @@ -12,7 +12,9 @@ def load_qwen_nunchaku(repo_id): shared.log.error(f'Load module: quant=Nunchaku module=transformer repo="{repo_id}" low nunchaku version') return None if repo_id.lower().endswith('qwen-image'): - nunchaku_repo = f"nunchaku-tech/nunchaku-qwen-image/svdq-{nunchaku_precision}_r128-qwen-image.safetensors" # r32 vs R128 + nunchaku_repo = f"nunchaku-tech/nunchaku-qwen-image/svdq-{nunchaku_precision}_r128-qwen-image.safetensors" # r32 vs r128 + elif repo_id.lower().endswith('qwen-lightning'): + nunchaku_repo = f"nunchaku-tech/nunchaku-qwen-image/svdq-{nunchaku_precision}_r128-qwen-image-lightningv1.1-8steps.safetensors" # 8-step variant else: shared.log.error(f'Load module: quant=Nunchaku module=transformer repo="{repo_id}" unsupported') if nunchaku_repo is not None: