JuliaWolleb

Hi In the settings here, the model will be saved in all 10,000 steps (iterations). For my application, around 80,000 iterations were enough (with a batch size of 10) ....

Hi I checked the numbers for training and testing and you are right, there is a typo: In the training set, we have 5,598 healthy slices, and 10,670 (instead of...

Hi Not using Visdom does not affect model training. this is just a visualization tool. Yes, you need the classifier as well as the diffusion model to train on both...

Eh, so what was the goal of your task? I think this data should be suitable, if you change from one class to the other? did you try to tune...

I see. You need two groups of images for your training, one only containing clean signals and the other one containing defected signals. Then the classifier should provide gradient maps...

Hi As pointed out in the paper, we trained the DDPM model for 50,000 iterations. This took roughly 2 days. The meaning of ema is "exponential moving average". I picked...

Hi How did you set the path to your training data? It should go to _data/chexpert/training._ Then, the length of the class names should be 6 instead of 8, because...

yes, one .npy file represents one image. Would maybe be a better idea to store them in another format. What do you mean by transfer code?

Hi Yes this is normal and due to the fact that the input images contain various levels of noise.

That is usually hard to tell. You can validate by computing the classification scores on non-noisy images.