Welcome to BudgetML Discussions!

[htahir1]

👋 Welcome!

We’re using Discussions as a place to connect with other members of our community. We hope that you:

• Ask questions you’re wondering about.
• Share ideas.
• Engage with other community members.
• Welcome others and are open-minded. Remember that this is a community we
  build together 💪.

To get started, comment below with an introduction of yourself and tell us about what you do with this community.

[JulesBelveze]

Hi @htahir1 , thanks for the super handy library !

I am wondering whether or not it is possible to include some extra python file when creating the Docker container?
I am attempting to infer a custom model and thus I need a bunch of files like: checkpoint, model file, config and so on..
I couldn't find anything mentioning this in the docs.

Thanks for your help :)

[htahir1]

Hi @JulesBelveze ! Currently, it is not possible to do this: The way it works is that the container itself does not contain the model or any data - just the environment dependencies and server code. That said, you could always create your own container and pass it in the launch function with the dockerfile_path parameter, if you wanted custom stuff in it. You can then make a copy of the server and add whatever you like in there. Maybe thats a route you might want to explore? Let me know if you need help there.

Another way to do it is to actually download the checkpoint, model_file, config etc that you need in your Predictor's load method and then use it later on. You can easily put all the files you need in a google cloud bucket in the same project and download them, just like the actual predictor_class is downloaded in the server.

Yet another way would be to make modifications to the library itself and that would be more involved. The part which would change is the uploading of the predictor_class to the Bucket and then downloading it subsequently in the startup script of the instance: Most of the code is here. Perhaps you could contribute to the repo with your own design -> Super happy for your contributions!

Hope any/all of this helps! Let me know if you have any further issues -> If you do please feel free to open an issue rather than a discussion as that seems to more fitting on the issue board

[JulesBelveze]

Thanks a lot for your quick answer @htahir1 !
I will look into your suggestions !

Will open an issue right away :)