nicojahn / Determining-the-Relevance-of-Features-for-Deep-Neural-Networks Public

Notifications You must be signed in to change notification settings
Fork 1
Star 0

WIP: Implementation of "Determining the Relevance of Features for Deep Neural Networks" by Reimers et al. (2020)

BSD-3-Clause license

0 stars 1 fork Branches Tags Activity

Name		Name	Last commit message	Last commit date
Latest commit History 2 Commits
ML_GCN @ 5517d6f		ML_GCN @ 5517d6f
.gitignore		.gitignore
.gitmodules		.gitmodules
Custom.ipynb		Custom.ipynb
LICENSE.md		LICENSE.md
MS_COCO.ipynb		MS_COCO.ipynb
README.md		README.md

Repository files navigation

My take on evaluating ...

"Determining the Relevance of Features for Deep Neural Networks", Reimers et al. (2020)

Each notebook should be rather self-contained (maybe 1-2 missing packages)
At the end of each notebook are open questions

Get the submodule (for the MS COCO example)

git submodule update --init --recursive

Download all complementary data (for the MS COCO example)

get the coco model: https://drive.google.com/open?id=1ivLi1Rc-dCUmN1ProcMk76zxF1DSvlIk
put it into: ML_GCN/checkpoint/coco/coco_checkpoint.pth.tar
get the valdiation data: http:https://images.cocodataset.org/zips/val2014.zip
get the annotations: http:https://images.cocodataset.org/annotations/annotations_trainval2014.zip
place both files into the directory: ML_GCN/data/coco/tmp/