Next steps

[ineiti]

Here some jumbled possible next steps - to be cleaned up as we go along:

• create our own testing data using StableDiffusion (mostly if we see we don't have enough with the MVTech data)
• implement the slicing of the ResNet intermediary layer: instead of having the whole layer inputted to the MLP, do a sliced implementation like in PatchCore and PNI, so that we can know where the error is

[lanterno]

We should close this ticket.