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鈥檒l occasionally send you account related emails.

Already on GitHub? Sign in to your account

0.1.0: Backend and Front-end refactoring, UI revamp #6

Merged
merged 4 commits into from Mar 25, 2024

Conversation

matteocargnelutti
Copy link
Collaborator

@matteocargnelutti matteocargnelutti commented Mar 20, 2024

This PR is a candidate for a v0.1.0 of WARC-GPT, which now looks like this:

Screen.Recording.2024-03-20.at.5.09.28.PM.mov

It mainly consists in:

  • A code refactoring of the backend: Departing from LiteLLM, decoupling search from text completion.
  • A code refactoring of the front-end: Bring some modularity through web components to make the front-end easier to edit and extend.
  • A UI revamp: Focus on a more "chatbot like" UI

A few other things that were added / removed:

  • Added: Streaming of text completion. Thanks to @edsu for kicking-off that work!
  • Added: Ability to configure endpoint-specific rate limits
  • Added: Session logs now also print in the browser's console and in an "Inspect" window
  • Removed: The ability to replace RAG prompt on the fly

Many thanks in advance to @rebeccacremona who will be having a look at this PR 馃槃

See #6 for details
gpt-3.5 series models can also be used in that context.
Copy link

@rebeccacremona rebeccacremona left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Wow, this is a huge PR!

Looks good to me. I read everything through and did not spot anything; I did not try running it locally.

I left a couple very tiny suggestions, all of which can certainly be ignored.

Re: the new UI, I'll note that the streaming text and auto-scrolling is a bit too much motion for me personally... it might be nice to have an option that let a user toggle that off for a steadier experience, maybe off if the user prefers-reduced-motion? But that sounds like something that could be tacked on later; I'm not suggesting it block this PR!!

README.md Show resolved Hide resolved
warc_gpt/static/components/ChatBubble.js Show resolved Hide resolved
warc_gpt/static/components/ChatFlow.js Outdated Show resolved Hide resolved
warc_gpt/static/components/ChatInput.js Outdated Show resolved Hide resolved
warc_gpt/static/components/InspectDialog.js Show resolved Hide resolved
warc_gpt/static/components/SettingsDialog.js Outdated Show resolved Hide resolved
warc_gpt/utils/list_available_models.py Show resolved Hide resolved
warc_gpt/utils/list_available_models.py Outdated Show resolved Hide resolved
- Added support for prefers-reduced-motion (no auto scroll)
- Updated "Enforce singleton" pattern
- Put trace logs in logger.debug calls
- Updated instructions of OPENAI_BASE_URL
@matteocargnelutti
Copy link
Collaborator Author

matteocargnelutti commented Mar 25, 2024

Thank you so much for having a look @rebeccacremona!

it might be nice to have an option that let a user toggle that off for a steadier experience, maybe off if the user prefers-reduced-motion? But that sounds like something that could be tacked on later; I'm not suggesting it block this PR!!

Definitely worth doing: I added logic to disable auto-scroll when prefers-reduced-motion: reduce is detected.

@matteocargnelutti matteocargnelutti merged commit 78a29db into main Mar 25, 2024
@matteocargnelutti matteocargnelutti deleted the refactoring branch March 25, 2024 22:14
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

2 participants