Daniel Falbel

For some reason I can't run the tests nbtests locally. They just seem to be hanging and never finish: ``` go test ./internal/... -run ./internal/nbtests/dom_test.go ? github.com/janpfeifer/gonb/internal/comms [no test files]...

Thanks @s-u You are right, we should definitely have a single liner `cmake` call that builds both lantern and libtorch. Maybe adding another option that uses cmake's [`ExternalProject`](https://cmake.org/cmake/help/latest/module/ExternalProject.html) module and...

Indeed we own a lot of improvements regarding documentation for package contributors. We do have an article here: https://torch.mlverse.org/docs/articles/modifying-source-code But for some reason it didn't get indexed in the website...

I think I know what the issue is. Pandas is discovered when ipykernel is first loaded, if we install pandas within the current session (eg with %pip install pandas) then...

I don't see any other reasons for that to happen. Can you try in the smoket ests if running `%view df` returns an error?

I ended up approaching this by forwarding the error to the user instead of lazily loading pandas as it's a simpler change and also solves other possible issues during initialization...

I agree with your second suggestion `[` should ibnclude the batch dimension when called with a single index on a dataset that only implements `.getitem()`. We could implement `[[` to...

Yes, maybe a simpler solution is to error if it only implements `.getitem()` but then, I don't think we should include the batch dimension in this case. Maybe just allow...

I heavily recommend using the pre-built binaries from: https://torch.mlverse.org/docs/articles/installation#pre-built The pre-built binaries bundle the necessary CUDA and cudnn versions, so you don't need a global compatible system version of CUDA.

Did you run something like: ``` options(timeout = 600) # increasing timeout is recommended since we will be downloading a 2GB file. kind