Views
No views yet
alexzhou907/DDBM.pytorch-image-translation-models, not the standard DDBMPipeline.from_pretrained.| Model variant | Domain |
|---|---|
edges2handbags-vp | Edges -> Handbags |
diode-vp | DIODE image translation |
1DDBM-ckpt/
2 edges2handbags-vp/
3 unet/
4 config.json
5 diffusion_pytorch_model.safetensors
6 diode-vp/
7 unet/
8 config.json
9 diffusion_pytorch_model.safetensors1from examples.community.ddbm import load_ddbm_community_pipeline
2
3pipe = load_ddbm_community_pipeline(
4 "/path/to/DDBM-ckpt/edges2handbags-vp",
5 device="cuda",
6)
7
8source = ... # PIL Image or torch.Tensor
9out = pipe(source_image=source, num_inference_steps=40, output_type="pil")
10out.images[0].save("ddbm_output.png")pytorch-image-translation-models with the community DDBM package..pt checkpoints, convert to unet/ format:1python -m examples.community.ddbm.convert_pt_to_unet \
2 /path/to/DDBM-ckpt/edges2handbags-vp \
3 --checkpoint e2h_ema_0.9999_420000.pt1@inproceedings{zhou2024ddbm,
2 title={Denoising Diffusion Bridge Models},
3 author={Zhou, Linqi and Lou, Aaron and Khanna, Samar and Ermon, Stefano},
4 booktitle={ICLR},
5 year={2024}
6}