continue icon indicating copy to clipboard operation
continue copied to clipboard

Published 20 hours ago • verified publisher icon continuedev

Integrating PiecesOS LLM in Continue

Open sambhavnoobcoder opened this issue 1 year ago • 3 comments

Description

Added support for Pieces OS as a new LLM provider in Continue. This integration allows users to utilize Pieces OS for completions and other supported operations within the Continue environment.

Checklist

  • [ ] The base branch of this PR is dev, rather than main
  • [x] The relevant docs have been created and updated

Details of Changes

  1. Created piecesOSLLM.ts in the core/llm/llms directory
  2. Implemented PiecesOSLLM class extending BaseLLM
  3. Set up PiecesOS provider and client
  4. Implemented key methods: _convertArgs, _streamComplete, _call
  5. Added getters for LLM type and identifying parameters
  6. Updated index.ts to include PiecesOS
  7. Updated config_schema.json for PiecesOSLLM configuration
  8. Created piecesOSLLM.md documentation
  9. Added setup instructions, configuration details, usage guidelines, and limitations

sambhavnoobcoder avatar Sep 17 '24 09:09 sambhavnoobcoder

Deploy Preview for continuedev failed. Why did it fail? →

Name Link
Latest commit 56a5ad4fe319bbe649f9fcb16fa79c26ea153ea8
Latest deploy log https://app.netlify.com/sites/continuedev/deploys/66e9c558e72a020008ae5bfa

netlify[bot] avatar Sep 17 '24 09:09 netlify[bot]

@sestinj could you share your thoughts regarding this

shivay-couchbase avatar Sep 17 '24 10:09 shivay-couchbase

Hi @sambhavnoobcoder—sorry for the delay. I'm looking into Pieces OS more and thinking that something even better than this PR might be OpenAI-compatibility so that a custom provider isn't needed. We're reaching a point where there are such a large number of LLM providers to maintain that it is likely to cause many of them to fall out of regular maintenance. The more providers can use OpenAI-compatible servers, the more stable and up-to-date support we can offer for them, and I'd strongly prefer to do this for Pieces.

Is Pieces OS OpenAI-compatible, or is this a change that could be made? I would estimate that this would require fewer lines of code within PiecesOS than it would in this current PR

sestinj avatar Sep 29 '24 02:09 sestinj

Hi @sambhavnoobcoder, since it's been a while and I think that the OpenAI-compatibility approach would be much better, I'm going to close this PR for the time being. Happy to take another look if you have the chance to look into the OpenAI-compatible route

sestinj avatar Oct 13 '24 06:10 sestinj