Instructions to use OPPOer/Qwen-Image-Pruning with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use OPPOer/Qwen-Image-Pruning with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("OPPOer/Qwen-Image-Pruning", dtype=torch.bfloat16, device_map="cuda") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- Draw Things
- DiffusionBee
Which blocks were removed?
This is amazing!. If you don't mind me asking, which 20 blocks were removed from the parent model? Was student teacher training done on the whole model, or just the blocks near the pruned blocks?
Thank you for your attention, Ostris. We mainly identified some prunable layers through layer sensitivity analysis and then conducted alignment training. We will release the detailed technical report in the near future. Actually, I have been following your works, from your early OpenFlux to the recent Qwen-Image-Edit-Inpainting project—all are excellent open-source initiatives.
Thank you for your attention, Ostris. We mainly identified some prunable layers through layer sensitivity analysis and then conducted alignment training. We will release the detailed technical report in the near future. Actually, I have been following your works, from your early OpenFlux to the recent Qwen-Image-Edit-Inpainting project—all are excellent open-source initiatives.
Any timeline? Look forward to it!
Our preliminary plan is to release it around November