Feature/Mixtral test

[EstebanDalelR]

Description

Proposal to use Mixtral on Together AI

Type of change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Documentation update
• Chore: cleanup/renaming, etc
• RFC
• Test

Notes

Acceptance

• I have read the Contributing guidelines
• I have read the Code of Conduct

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Comments	Updated (UTC)
watermelon	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	Dec 14, 2023 1:56am

[watermelon-copilot-for-code-review]

Watermelon AI Summary

This pull request introduces a new feature that replaces OpenAI with Mixtral in the Together AI system, adding the necessary integration files and tests for the newly proposed functionality.

GitHub PRs

• #414 - Feature/Mixtral test - By EstebanDalelR Just now

• #412 - Test/Add Test File - By baristaGeek 1 days ago

• #399 - Test/pingign route - By EstebanDalelR 1 weeks ago

• #413 - Feature/better console log detection - By baristaGeek 6 hours ago

• #62 - Chore/remove supporters section in readme - By baristaGeek 1 months ago

Click here to login to Jira
Click here to login to Confluence
No results found in Slack Threads :(

Notion Pages

• #undefined - Write project proposal

No results found in Linear Tickets :(

Click here to login to Asana
watermelon is an open repo and Watermelon will serve it for free.
🍉🫶

[watermelon-copilot-for-code-review]

This PR contains console logs. Please review or remove them.

[baristaGeek]

Code looks fine and works fine. I'll approve. We can go ahead and add Mixtral to the other endpoints.

However, I do want to approve with this in mind. This is the Mixtral result for #412:

And this is the one for Open AI:

Open AI's result has higher quality. Is it just a matter of time that Mixtral's quality gets there?

[EstebanDalelR]

Code looks fine and works fine. I'll approve. We can go ahead and add Mixtral to the other endpoints.

However, I do want to approve with this in mind. This is the Mixtral result for #412:

And this is the one for Open AI:

Open AI's result has higher quality. Is it just a matter of time that Mixtral's quality gets there?

I think it's 2 things:

• Mixtral is a GPT3.5 level model
• and the one we are using is not chat optimized, so we need a better prompt.

On the other hand, this PR proves that it works! We can now use Together to deploy pretty much any OSS model.

My proposal is to make GPT the default and add a random text suggesting you try others?

[EstebanDalelR]

Not yet ready for production.