GenerativeModels
GenerativeModels copied to clipboard
Project-MONAI
MONAI Generative Models makes it easy to train, evaluate, and deploy generative models and related applications
Add components necessary to perform image to image translation (including sampling process condintioned on images) based on Generative Adversarial Networks.
Add components necessary to perform image to image translation (including sampling process condintioned on images) based on diffusion models.
If the dimension k=3 for three dimensional data, then the following -1 should be -3/2? `line 259: kl_loss = 0.5 * torch.sum(z_mu.pow(2) + z_sigma.pow(2) - torch.log(z_sigma.pow(2)) - 1, dim=[1, 2,...
I am trying to train a AutoEncoderKL model on RGB images with the following dimensions (3,1225,966). Here is the code that I use ( similar to what's there in tutorials/generative/2d_ldm/2d_ldm_tutorial.ipynb...
After wrapping a MONAI AutoencoderKL model in pytorch lightning, I got an error regarding an unused parameter in the custom AttentionBlock implementation. [diffusion_model_unet.py](https://github.com/Project-MONAI/GenerativeModels/blob/main/generative/networks/nets/diffusion_model_unet.py) and [autoencoderkl.py](https://github.com/Project-MONAI/GenerativeModels/blob/main/generative/networks/nets/autoencoderkl.py) line 233: ``` self.proj_attn =...
Well, the reversed_step function in generative.networks.schedulers.ddim.py seems to try to generate x_t+1 from x_t with the Equation (6) in https://arxiv.org/pdf/2203.04306.pdf. However, in the process of implementing the reversed_step function, from...
whenever context is passed to a block like in: h, res_samples = downsample_block(hidden_states=h, temb=emb, context=context) the forward function uses "del context" so it is not really implementing this conditioning. further...
Metadata
Owner
Metadata
MONAI Generative Models makes it easy to train, evaluate, and deploy generative models and related applications