Add OpenAI provider about twinny

[rjmacarthy]

Hello, you can use OpenAI by using the LiteLLM provider proxy.

[walkandre]

Just add a new provider with the settings:

[JamesClarke7283]

Oh, thanks, i will keep that in mind.
I did not see OpenAI anywhere in the docs, so thats why i could not find out. But thanks/

[bareblackfoot]

@walkandre where is this setting in the vscode?

[rjmacarthy]

@walkandre where is this setting in the vscode?

@bareblackfoot

You should click the plug 🔌 icon at the top. You should also install a local proxy for OpenAI such as LiteLLM or similar.

[bareblackfoot]

@rjmacarthy
Thanks for your fast response! I found the button and it worked well. The chat model works fine. However, the gpt3.5-turbo-instruct model for FIM generates <PRE> and <MID> tokens. Do you know how can i fix it?

Providers
Label*: OpenAI FIM
Type*: fim
Fim Template*: automatic
Provider*: litellm
Protocol*: http
Model name*: gpt-3.5-turbo-instruct
Hostname*: 0.0.0.0
Port*: 34162
API path*: /chat/completions

[rjmacarthy]

@rjmacarthy Thanks for your fast response! I found the button and it worked well. The chat model works fine. However, the gpt3.5-turbo-instruct model for FIM generates

 and  tokens. Do you know how can i fix it?

Providers
Label*: OpenAI FIM
Type*: fim
Fim Template*: automatic
Provider*: litellm
Protocol*: http
Model name*: gpt-3.5-turbo-instruct
Hostname*: 0.0.0.0
Port*: 34162
API path*: /chat/completions

Hey I'm not sure about FIM with OpenAI API...I saw some legacy stuff before but never tested it. You could try a custom FIM template or explore the API docs. Please let me know how you get on.