Pre-trained models?

[anguyen8]

Dear authors,

Thank you so much for releasing the code of this fascinating work!
Is there any chance you could also release the pre-trained ProtoPNet models used in your paper?

This would be very useful for us and the community, I believe. Thanks!

Anh

[Dadatata-JZ]

I would like to second this suggestion. I followed the instruction and the result is not reproducible (indeed, the testing accuracy I got was way off). Is there any tips for training protoPnet? @cfchen-duke Thank you in advance.

[anaghajoshi111]

I would like to second this suggestion. I followed the instruction and the result is not reproducible (indeed, the testing accuracy I got was way off). Is there any tips for training protoPnet? @cfchen-duke Thank you in advance.

@Dadatata-JZ Can you tell me what versions of PyTorch, NumPy, cv2, and Augmentor you used to get the accuracy?

[KodeRikke]

@Dadatata-JZ Did you figure out this issue? We are having trouble reproducing the accuracy too.