Can't get notebook to work on Windos 11

[yulworm]

Not sure if this is the best venue for this, but I am trying to run your notebook on my Windows 11 pc and it never gets past:
self.db = vectorstores.Chroma.from_documents(self.texts, self.select_embedding)

• running python 3.12.0
• 5gb of free RAM
• trying to load a small (4 page) document

It might be that this will not run on Windows; I had to make some changes to the requirements to get pip to finish:

• removed uvloop since it is not supported on windows.
• typer - removed any fixed version
• onnxruntime==1.17.3 instead of 1.17.1

My problem looks like https://stackoverflow.com/questions/77093497/can-not-run-langchain-module-kernel-died-while-running-the-script, but I would rather not have to use Colab

Great talk.
Thanks for any help.

[t-redactyl]

Thanks for reaching out, and sorry you're having troubles running the notebook. Unfortunately I haven't been a Windows user for years (I actually moved away from Windows due to issues like this with Python!), so I'm probably not going to be able to help debug the issues with running the code directly on Windows.

I'd say the easiest fix is going to be using a Linux machine: if you don't want to use Colab, you can also use WSL or run the notebooks in a Docker container. For WSL, I wrote a guide on how to use this in PyCharm (you can of course find similar guides for other IDEs if you're not a PyCharm user): https://blog.jetbrains.com/dataspell/2022/07/using-jupyter-notebooks-with-wsl-2-in-dataspell/ (I wrote it for DataSpell but it works the same in PyCharm).

Otherwise, you can run a Jupyter notebook from a Docker container. Here is the Dockerfile I normally use to run my notebooks:

FROM python:3.11.7 # or change to your desired version

RUN pip install notebook [project dependencies here] && mkdir /notebooks

WORKDIR /notebooks

ENTRYPOINT ["jupyter-notebook","--no-browser","--allow-root","--ip","0.0.0.0"]

Once you've created the Dockerfile, build and run it in the usual way by running the following commands in the same directory as the Dockerfile:
docker build . -t rag/jupyter
docker run -p 8888:8888 -v ./notebooks:/notebooks rag/jupyter

Then copy the URL starting with 127.0.0.1:8888 into the browser and your notebook is ready to use.

Hope this helps!