Closed vsyrgkanis closed 2 months ago
Hi @vsyrgkanis thanks for all of these changes. I am in the process of reviewing them. Currently I checked all notebooks AC, CM and PM1 and PM2.
I checked all generated Rmds in branch gen-Rmd
. The transformation from irnb to Rmd seems to work very well. I was able to execute and render them. The results seem to be consistent with everything before.
Also the liniting improves the coding style nicely. I found one notebook where the linting had unwanted results:
The linting issues I could correct in the .irnb files so that next time the Rmds are generated they are fine too. However, they might be linted back next time the lint action runs.
Maybe a small remark, if you render the Rmd to html, the title is "An R Markdown document converted from ..." which is maybe not super pretty but also not really an issue.
I will check all upcoming notebooks as soon as they finished rendering on my machine. Is there anything else you would like me to look at?
Best, Oliver
Sorry for the reverted commit. Got confused since execution failed both locally and on colab and only then saw that keras < 3.0.0 is a dependency here.
I was able to render all the Notebooks now and they look fine and are consistent with the irnb.
Sorry for the reverted commit. Got confused since execution failed both locally and on colab and only then saw that keras < 3.0.0 is a dependency here.
I was able to render all the Notebooks now and they look fine and are consistent with the irnb.
Yes there was an issue with our code with Keras3 and beyond, so had to add this dependency. Otherwise we need to change a lot in our code, if we want to support keras3.
Also @OliverSchacht @MartinSpindler one protocol is that approvers, just add comments and request to a PR and not directly make changes and pushes!
This PR makes the following changes: