YashNagraj75/Latent-Diffusion-Conditional

Add model card

Add wandb logging (Life is more peacefull)

Model not moving forward

Add update logic

Add scheduler

Sloving conflicts

More stuff

Update disc_loss update logic

Made changes in the training script

More samples

Update

Re training will finish this time for sure

More samples

Halfway through

Remove all samples

Checkpoint for 2nd epoch

Checkpointnt for 1st epoch

Change checkpoint dir

Modify training code

Remove logging statements

Move to gpu and add checkpoint images to attributes

Print statements for logging

Add num_heads variables

Make changes

Resolve run time errors

Make changes to get image

Get the buffers in order to open as image

Change any output errors

Add im_path from the config

Change the config file accordingly

Add loss printing and saving logic

Optimize Discriminator in the training loop

Change dataloader loading from directory

Add dataloader for training

Add the dataset

Move the pretrained vgg16 model to another directory

Add forward function to the Diffusion model UNet

Build the Unet from the config

change variable names

Get all parameters from model_config

Add discriminator for adverserial loss for VAE training

Add lpips for VAE training

Clean up

Add AutoEncoder (VQVAE)

Add UpBlock Unet to concat downblock's output

Add all the layers in the UpBlock

Add forward function to the midblock

Add residual convolution to the mid block

Add MidBlocks and change cross_attn in down blocks

Add cross attn if needed for conditional latent diffusion