Open gg22mm opened 1 month ago
Thank you for the project, it's definitely worth learning from~~ / 感谢大佬的项目,非常值得学习 ~~!
大老: 测试中发现这些问题,其它的都可以跑通没问题 / Da Lao: These issues were found during testing, but everything else can run smoothly without any problems ~~
我发现: Super-resolution超分辨率 少了数据集调有代码 / Old man: Super resolution is missing, but there is code for adjusting the dataset: https://github.com/CreamyLong/stable-diffusion/blob/master/configs/retrieval-augmented-diffusion/768x768.yaml
768x768.yaml no data and lightning
数据集用的是 / What is the dataset used for: https://www.kaggle.com/datasets/wangrui071/mscoco-2017-trainval-annotations python mail.py --base configs/latent-diffusion/layout2img/coco_sg2im_ldm_Layout2I_vqgan_f8.yaml --train True
'''# 这里有问题 self.transformer_blocks = nn.ModuleList( [BasicTransformerBlock(inner_dim, n_heads, d_head, dropout=dropout, context_dim=context_dim[d], disable_self_attn=disable_self_attn, checkpoint=use_checkpoint) for d in range(depth)] )'''
#wll 修改 context_dim=context_dim[d] 改成 context_dim=context_dim[0],这样是没报错了,但是不知道有没有别的问题,也有可能是我的数据集有问题,只否提供一个测试数据集呢? self.transformer_blocks = nn.ModuleList( [BasicTransformerBlock(inner_dim, n_heads, d_head, dropout=dropout, context_dim=context_dim[0], disable_self_attn=disable_self_attn, checkpoint=use_checkpoint) for d in range(depth)] )
python main.py --base configs/latent-diffusion/mask2img/config.yaml --train True
少了:flickr 类 / Missing: Flickr class
Super-resolution
Depth-to-Image. Currently, I may not have time to implement it
Thank you for the project, it's definitely worth learning from~~ / 感谢大佬的项目,非常值得学习 ~~!
大老: 测试中发现这些问题,其它的都可以跑通没问题 / Da Lao: These issues were found during testing, but everything else can run smoothly without any problems ~~
1 Super-resolution(超分辨率)
我发现: Super-resolution超分辨率 少了数据集调有代码 / Old man: Super resolution is missing, but there is code for adjusting the dataset: https://github.com/CreamyLong/stable-diffusion/blob/master/configs/retrieval-augmented-diffusion/768x768.yaml
768x768.yaml no data and lightning
2 Layout-to-Image(布局到图像):
数据集用的是 / What is the dataset used for: https://www.kaggle.com/datasets/wangrui071/mscoco-2017-trainval-annotations python mail.py --base configs/latent-diffusion/layout2img/coco_sg2im_ldm_Layout2I_vqgan_f8.yaml --train True
'''# 这里有问题 self.transformer_blocks = nn.ModuleList( [BasicTransformerBlock(inner_dim, n_heads, d_head, dropout=dropout, context_dim=context_dim[d], disable_self_attn=disable_self_attn, checkpoint=use_checkpoint) for d in range(depth)] )'''
3 Semantic-to-Image(语义到图像):
python main.py --base configs/latent-diffusion/mask2img/config.yaml --train True
![图片](https://github.com/user-attachments/assets/42317f30-5a66-47aa-b794-75834ab5b77a)
少了:flickr 类 / Missing: Flickr class
4 Depth-to-Image(图像深度 , 未开发 Image depth, undeveloped)