| --- |
| license: apache-2.0 |
| base_model: |
| - Qwen/Qwen-Image-2512 |
| - Qwen/Qwen2.5-VL-7B-Instruct |
| --- |
| |
| Update 19/02/26: uploaded v2 with some layers kept at BF16 and longer calibration (more steps, adjusted learning rate, etc). Clearly better than the first version I did, doesn't get "fuzzy" on edges or hair. |
|
|
| -------- |
|
|
| NVFP4 version of [Qwen-Image-2512](https://huggingface.co/Qwen/Qwen-Image-2512) and [Qwen2.5-VL-7B-Instruct](https://huggingface.co/Qwen/Qwen2.5-VL-7B-Instruct), both created from BF16 versions in [Comfy-Org/Qwen](https://huggingface.co/Comfy-Org/Qwen-Image_ComfyUI) repo. Made with [Silveroxides/convert_to_quant](https://github.com/silveroxides/convert_to_quant) script. |
| Drop-in replacement in Comfy, there's a bit of quality loss as in nunchaku models or others nvfp4. |
|
|
| Conversion command: |
| `convert_to_quant -i qwen_image_2512_bf16.safetensors -o qwen-image2512-nvfp4.safetensors --nvfp4 --comfy_quant --qwen` |
|
|
| ⚠️ I strongly suspect a bottleneck somewhere (memory bandwidth? dequant ops?) that throttles the GPU down in some cases, running small batches instead of single image seems to remove it. I only have surface knowledge of this stuff and don't know how to troubleshoot it. |
|
|
| | Batch of 4 | Qwen-Image-2512-fp8 | Qwen-Image-2512-nvfp4 | |
| |------------|------------------------|------------------------| |
| | Vanilla | 194s (45.6s per image) | 76s (18.3s per image) | |
| | Lightning | 20.4s (5s per image) | 7.5s (1.75s per image) | |
|
|
| (wall time, on a 5090. Vanilla = 50 steps & CFG 4, Lightning = 8 steps & CFG 1 with [Lightning lora](https://huggingface.co/lightx2v/Qwen-Image-2512-Lightning)) |
