Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[v2][Enhancement] Improve project and code documentation using LLMs #207

Open
Udayraj123 opened this issue Sep 15, 2024 · 12 comments
Open
Assignees

Comments

@Udayraj123
Copy link
Owner

Udayraj123 commented Sep 15, 2024

Is your feature request related to a problem? Please describe.
LLMs like ChatGPT can help make our project documentation easier to understand and fill any missing gaps.

  • Currently the project documentation is present in the project wiki.
  • Also the code comments are scarce except for the variable naming and the descriptions present json schemas that give some context.

Describe the solution you'd like
We should leverage code understanding capabilities of the LLMs to greatly improve documentation. We can -

  • Improve the existing User Guides in the Project wiki
  • Incrementally add explanatory code comments which can be read by docs generators later
    • Looking for a good prompt that does this job well

Describe alternatives you've considered
Developers have to refer to the provided samples to learn how to use each of the flags present in our template/evaluation schemas

Additional context
We are also planning to move it into a dedicated docs/pages folder in future. Also exploring using docs generators like Docusaurus for more elaborate project and code documentation.

Note: please share your queries and approaches on discord for quicker discussions:

@chirag4862
Copy link

I would love to work on this enhancement.

@Udayraj123
Copy link
Owner Author

Hey @chirag4862, good to hear that! Docs are crucial for a good adoption of the OMRChecker and I hope you'll be able to help them improve. Please go through the current wiki and share which part of it you'd like to work on.

@offline-keshav
Copy link

offline-keshav commented Oct 2, 2024

@Udayraj123 we can improve documentation by following steps

  1. We need to improve explained comments in every program and module.
  2. After that we need to improve 2 instructions manuals , one for normal user and other for devs.
  3. The dev manual need to have working of project , which program does what. by this solving other issues will be very easy since file can be found/ accessed more easily and direct references to the files can be given.
  4. LLMs like GPT can help with all these. I have done this thing in my earlier project too which can be accessed here and it did it very efficiently though there were some changes that need to be made.

Also I would like to work on this issue since.

@Udayraj123
Copy link
Owner Author

This is awesome. I'm assigning the task to you. Let me know if you need any clarification. Let's do it file by file I suppose. One more pointer is the schema/code differences in dev vs master branch. I think you can proceed by documenting the dev branch code since I will be releasing the v2 soon.

@Udayraj123 Udayraj123 changed the title [Enhancement] Improve project and code documentation using LLMs [v2][Enhancement] Improve project and code documentation using LLMs Oct 2, 2024
@Udayraj123
Copy link
Owner Author

Unassigning due to unavailability, discussed in discord

@fa-anony-mous
Copy link

Hi Uday, I would like to work on this project where I would like to supply the files to a GPT model along with what each file does.
Then make it add comments wherever necessary, this can also be done in real time if I deploy the model with the workflow using lang graph. I would like to know your insights and feedback.

@Udayraj123
Copy link
Owner Author

hey @fa-anony-mous, that looks like an interesting approach. I'll assign the task to you and you can host sample docs in your fork of the OMRChecker repository. Based on how it looks we can proceed with any review changes if required

@fa-anony-mous
Copy link

fa-anony-mous commented Jan 30, 2025

thank you @Udayraj123 . As more code gets pushed into it, I would like to add a feature where you could go this application and ask the AI to refine or comment where necessary and later maybe you can look at it and review the changes. If that's the case, I will ask to enter google api studio key as their new model gemini 2.0 flash model has upto 1M tokens for free daily. I can add other options for LLM app keys as well. Let me know what do you think about this?

@Udayraj123
Copy link
Owner Author

Sure, we can discuss more on that later. It looks like a cool idea!

@fa-anony-mous
Copy link

hi @Udayraj123 I have added the docugen generator, i have tested out some outputs on some sample files. It works well.
Could you please check it out and try it yourself and let me know if you require anything else?

@Udayraj123
Copy link
Owner Author

@fa-anony-mous thanks, I'll try to check it by next week!

@Udayraj123
Copy link
Owner Author

Added a few comments. I'd appreciate if you could share the documentation output on OMRChecker using the PR you created.

Can you please share a sample somewhere in this PR or as an attachment?

Also we need to discuss and figure out how the documentation should update progressively on updating the source code(via new PRs)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

4 participants