Writing prompt

[fabraz]

This PR deliveries the code used to sample writing prompt dataset, which can be downloaded at https://huggingface.co/datasets/fabraz/writingPromptAug.

[andrewm4894]

@fabraz thanks!

can you add a readme in the folder and a open in colab button to the notebook.

Idea being someone can quickly scan the readme and then just open the notebook in Google colab even if they want to start playing with it.

https://github.com/LAION-AI/Open-Assistant/tree/main/notebooks/example

[fabraz]

@fabraz thanks!

can you add a readme in the folder and a open in colab button to the notebook.

Idea being someone can quickly scan the readme and then just open the notebook in Google colab even if they want to start playing with it.

https://github.com/LAION-AI/Open-Assistant/tree/main/notebooks/example

@andrewm4894, I just realized that this notebook would be better located under data-augmentation. Do you agree?

[andrewm4894]

Yep, maybe so. Kinda a lot of stuff ending up under data augmentation since it is very general but makes sense. Am sure some stage down the road we might reorder things in various ways but happy to keep it fairly organic for now.

So yep, maybe makes sense to add under data augmentation for consistency as of now.

[github-actions]

❌ pre-commit failed.
Please run pre-commit run --all-files locally and commit the changes.
Find more information in the repository's CONTRIBUTING.md

[fabraz]

Colab button and README.md cleared.

[andrewm4894]

notebook is looking a bit funny for some reason

seems like maybe some code cells and markdown getting messed up - is it just me?

https://github.com/fabraz/Open-Assistant/blob/writing_prompt/notebooks/data-augmentation/writing-prompt/writing_prompt.ipynb

[fabraz]

So weird. It seems that it converted all the cells into a single one. Checking …

[github-actions]

❌ pre-commit failed.
Please run pre-commit run --all-files locally and commit the changes.
Find more information in the repository's CONTRIBUTING.md

[fabraz]

@andrewm4894, I've found that the messy jupyter was due to a GitHub bug. When GH renders the following code, it lost any jupyter style setting thereafter.

I made some changes in the code, and now GH managed to present it correctly. Check it out.

I've done many commits for troubleshooting purposes. Let me know if it would be better to open another PR, to get rid of this commit messy history.

[andrewm4894]

oh interesting - ive seen markdown cells break github rendering before but never code cells lol

[andrewm4894]

i'm confused a bit by this - is that your gh username - im not sure if meant to be there or not?

[andrewm4894]

i also see an error so not all cells are rendered

ideally the notebooks would be a clean run top to bottom with all cells rendered so interested users can just read it before deciding if they want to play with it.

[fabraz]

@andrewm4894 I removed unnecessary comments and fixed the cell with error output. Check, please!

[olliestanley]

This should have had the .ipynb_checkpoints dir removed before merge really. Ah well

[andrewm4894]

#1358

Waiting to merge this