[SOLVED]I'm unable to run my model in production on Heroku because I made my model using kaggle

So for lesson 2 created an image classification model to distinguish between leopards and jaguars. I used 50 images of each and got 95% accuracy (notebook). I’m using the model and modifying the fast.ai render.com template and using flask instead (repository). I was able to get everything to work, including deployment to heroku thanks in part to this post. However, the application isn’t running and it seems to be because I created my model by using a kaggle notebook.


The kaggle/working line referenced in the error logs seems to be on line 187 of my export.pkl file.

I’d like to know if there’s anything I can do to fix this. Should I just try another site? For instance, what would be the costs for running this on render.com?

EDIT: I decided to just build the model locally instead of using the one I made on kaggle since I have the notebook anyway.

I had a same issue. Turns out you cannot write to kaggle as it is read only and that’s why the issue was happening. I used my own dateset and tried it again and it worked. You can check out my repo, it has the notebook in it to for details -

Hello Everyone,

I am getting issue as file size has increased the limit. Below is the error
! Compiled slug size: 1.1G is too large (max is 500M).