Cannot upload to the git even git lfs is checked ... windows using git tortoise

git.exe push --tags --progress “origin” master:master
Locking support detected on remote “origin”. Consider enabling it with:
git config lfs. true Locking support detected on remote "origin". Consider enabling it with: git config lfs. true
Enumerating objects: 56, done.
Counting objects: 100% (56/56), done.
Delta compression using up to 8 threads
Compressing objects: 100% (53/53), done.
remote: fatal: pack exceeds maximum allowed size
fatal: sha1 file ‘’ write error: Broken pipe
error: remote unpack failed: unpack-objects abnormal exit
error: failed to push some refs to ‘’

have you setup git-lfs properly? These are the steps I used:
(check that git-lfs is setup correctly in your git configuration files)

git lfs install

track the model files, if the models are saved as model.pth for example, use the following to track all pth files

git lfs track '*.pth'

Next, you need to add .gitattributes to your git repository.

git add .gitattributes 

Now that this is all set up, you should be able add the model to the git repo by

git add .
git commit -m "commit message"
git push -u origin master

Let me know if this helped.


Thanks Shraddha, my clients was not compatible with git lfs so i had to install new client … Thanks for the help though :slight_smile:

Screenshot from 2020-04-08 13-53-46
Screenshot from 2020-04-08 13-54-58
Screenshot from 2020-04-08 13-55-18
Screenshot from 2020-04-08 13-55-28
Screenshot from 2020-04-08 13-55-47

Screenshot from 2020-04-08 13-57-00
No models directory on gitlab repo. What could be the reason?

Can you check your .gitignore and see if .pth is mentioned there, if so remove it

1 Like

Thank you for the quick response!
Yes, .gitignore has had .pth in it.
I thought that we want it to have it because we only want LFS to handle it.
Now git push is uploading my model, thank you. I hope that now I will be able to create a legit submission.

1 Like

Do you have any idea what the problem could be?
@nikhil_rayaprolu pasted the error:

from skimage import measure
ModuleNotFoundError: No module named ‘skimage’

On my local machine I have:

import skimage

In the repository I’ve updated the requirements.txt so that it has the corresponding line:


And also the Dockerfile’s python3 pip3 install section to have the line:

What else should I do so that I could import it? Can you help me?

Clearly for some reason the module has not been installed in your docker image. I would suggest adding RUN pip install scikit-image in your Dockerfile. Since this is a docker issue, I think you could find the problem by debugging locally. It is explained in the baseline repo’s readme.

It seems you have a submission in progress already, I hope you’ve managed to solve your issue.

1 Like

Thank you very much for your response!
Yes, it seems that the evaluation is running now. At least it didn’t fail for 2 hours now. I hope that now I nailed it and so I can start to work on training the model.

Is that normal that after about 14 hours the submission still is running?

Not unless it would be something compute intensive. Have you managed to get your code running locally using nvdia-docker, if your code works locally on nvidia-docker then maybe it is a server issue.

Then I will have to dive into how docker works since I have never used it before and my submission failed half an hour ago.

C:\Users\lenovo\food-recognition-challenge-starter-kit>git push aicrowd master
Locking support detected on remote “aicrowd”. Consider enabling it with:
$ git config lfs. true
Connection to closed by remote host.

getting error like this.