cipriancorneanu / dnn-topology Goto Github PK

Dear Ciprian, your code looks amazing!

We are currently using it for some research but we found a line of code we can not understand. In this step of the code:

We do not get why you create two passers. We thought this could be related to reference passing and OOP stuff. We tried removing the "passer_test" variable and operating only with one passer and the results are the same. Is this an unintended code duplication or is it a bug?

Thank you for this interesting work!

Hi all. This is a good work but it seems that the README.md is out of date and not workable. Specifically,

The command to install DIPHA doesn't seem to work.

cd ./dipha
cmake -H. .Bbuild
cmake --build build -- -j3

and gives me

CMake Error: The source directory ".../dnn-topology/dipha/.Bbuild" does not exist.

Also it appears that the running file

python main.py --net lenet --dataset mnist --trial 0 --lr 0.001  --n_epochs_train 50 --epochs_test '10 20 30 40 50' --graph_type functional --homology_type 'persistent' --train 1 --build_graph 1

is not valid. There is no main.py file and the options do not work for train.py

Dear Ciprian,
first of all, thanks for this great project!

I think i found a bug when following the 'Quick Start'-section in the README.md. When running

python main.py --net lenet --dataset mnist --trial 0 --lr 0.0005  --n_epochs_train 50 --epochs_test '1 5 10 20 30 40 50' --graph_type functional --train 1 --build_graph 1

i receive this AttributeError

AttributeError: 'DataParallel' object has no attribute 'forward_features'

originating from the following line

I resolved it by replacing self.network.forward_features(inputs) with self.network.module.forward_features(inputs).

The problem seems to be that DataParallel does only forward specific function to the encapsulated model. This might also be a problem in other parts of the project (e.g. passers.py line 95).

Hello,

thank you for this project. I would like to use it and make some changes, but I wonder what is the license for this code? Could you please make that precise? Thank you in advance

Best wishes

Hi, I'll love to implement your approach to early stopping in a Keras environment. Have you've done this?