MaurerKrisztian/issue-improver-action

Allow usage for gpt-4 model

Closed this issue · 5 comments

kostecky commented

A possible enhancement could be to allow for gpt4 to be used. I tried specifying it as 'gpt4' and 'gpt-4', but it does not work. I assume it's because it needs to use the chatCompletion endpoint vs. createCompletion.

MaurerKrisztian commented

Yes, you are right. It has 2 endpoints with different models. Good idea.
https://platform.openai.com/docs/models/model-endpoint-compatibility

lduf commented

Any update here ?

MaurerKrisztian commented

@lduf I plan to make the necessary changes to this soon. Btw feel free to open a PR.

MaurerKrisztian commented

@lduf Done, GPT-4 now is available. To try out the latest features before the official release, use the latest flag: MaurerKrisztian/issue-improver-action@latest. You can specify the model input in the action but the default now is gpt-4.

Improvements

  • Updated OpenAI Library: Upgraded to the latest version of the OpenAI npm library, which includes API changes.
  • Endpoint Change: Transitioned from the legacy /v1/completions endpoint to the new /v1/chat/completions. For more details, see the Model Endpoint Compatibility documentation.
  • Configuration Options: Introduced an option in the configuration for system messages.
  • Code Refactoring: Eliminated redundant code for better efficiency and maintainability.

Pending Tasks (before release)

  • Update README: Modify README.md to reflect the latest changes. (systemMessage new config option, default model..)
  • Test: Test the current features
  • 🎉1
MaurerKrisztian commented

Done, new version v1.1 has been released.