Add dependency for ipykernel

[emptymalei]

ipykernel is required to create a new jupyter kernel.

Install dependencies

poetry install --with lab

Then

poetry run ipython kernel install --user --name=ts-diffusion-kernel

[cmp0xff]

Steps:

1. (Since I have already installed jupyterlab with conda in the ts-diffusion environment,) remove the conda environment ts-diffusion by conda env remove -n ts-diffusion
2. Install the environment following https://github.com/orgs/neuronstar/projects/2?pane=issue&itemId=19113095
   2.1. python==3.10.* is needed, otherwise no compatible torch version can be found.
   2.1. poetry and pre-commit install by brew
3. It turns out that poetry install already installs ipykernel, jupyter-core etc. Is this expected?

[emptymalei]

3 -> looks like ipykernel is already required by one of the existing packages.

I guess we just close this PR.

[cmp0xff]

3 -> looks like ipykernel is already required by one of the existing packages.

I guess we just close this PR.

I still need the package jupyterlab. Is it required by any other packages?

[emptymalei]

jupyterlab should be installed outside of the project, e.g., some jupyterlab env.

After poetry run ipython kernel install --user --name=ts-diffusion-kernel, jupyterlab will see the kernel ts-diffusion-kernel.

[cmp0xff]

Okay. What would be the process if I want to go back to main? Do I need to remove the conda environment and reinstall again?

[emptymalei]

My guess is poetry install --sync

https://python-poetry.org/docs/cli/#options-2

[cmp0xff]

• Updating watchdog (2.3.0 -> 2.2.1): Failed

Guess I have to uninstall the conda env...

BTW, the main branch does not contain ipython as a requirement.

[cmp0xff]

poetry run ipython kernel install --user --name=ts-diffusion-kernel

It does not work with the requirements in the main branch.

[emptymalei]

poetry run ipython kernel install --user --name=ts-diffusion-kernel

It does not work with the requirements in the main branch.

works for me

Is it a conda poetry compatibility problem? I am using virtualenv.

I can test on conda later.