few-shot-diffusion

few-shot-diffusion copied to clipboard

fusion model &ckpt

1

感谢您的工作！请问这两步中训练使用的数据集都是source data吗？  另外，我在加载您提供的训练好的checkpoint时，出现参数不匹配的情况。请问您使用的UNet网络是您现在代码所提供的吗？我不想重新进行训练，能否麻烦您帮助我解决一下这个问题？ 

bigwahaha

No such file or directory: 'features1.npy'

19

when i run the 'train-whole.py', an error occurs:'No such file or directory: 'features1.npy''. May I know how to solve it？

zhouzheng1123

"generate source-domain images" during inference

2

`If you hope to generate the target domain images, the simplest way is to generate source-domain images into $source_img_path, and then transfer these images to the target domain.` Can I...

fikry102

phasic content fusing model

3

`After the basic diffusion model is obtained, you can train phasic content fusing module by: python3 train-recon.py --data_path=$path_to_dataset --ckpt=$path_to_basic_diffusion_checkpoint ` So, we only use source dataset to train the phasic...

fikry102

a bit confused

1

Hi, could you explain the difference between a basic diffusion model and a phasic content fusing module? I'm a bit confused and would appreciate some clarification.

invade-art

SE module

1

thanks for your job,i have a question .In your SE module, did yt-1 import your model only once and generate y0?

1289600760

stop point "Train the whole model on both the source and target dataset"

3

`Once it is good enough, you can stop training. In our experience, for the style like cartoon and VanGogh_face, 300-500 iterations are enough while for Sketches, 3000 iterations work better.`...

fikry102

code for Quantitative Evaluation

1

`We first sample 1000 images from Style-GAN2 [15] as the source images and generate 1000 images in target domain by all the methods. Then we calculate the IS, IC-LPIPS and...

fikry102

ML-GCN

m(t)与w(t)

1

你好，我想问问sigmoid function m(t)和weighting function w(t)的具体公式是怎么想出来的啊?

15634960802