GPT 3.5 Turbo

[mangeloni30]

AI Data Dashboard

• Using gpt-3.5-turbo model

[mangeloni30]

Hey ! Thanks for the PR.

We need to make some changes in order to add this.

• First we should try to split the logic between the models. They behave differently, one returns texts and the other handles objects.
• For now let's keep it simple and let's keep the function generatePrompt returning a string for all models. So we can keep using the same UI for the prompt view.
• We should add a select input in the config panel (where you configure your API KEY) to be able to switch between models. This parameter should be saved and handled in the same way that the API KEY is being handled.
• For the functions handling the "chat/completions" endpoint, we shouldn't be doing any spliiting on the responses like we do for the completions, since the texts are already split into roles.

Let me know if you have any doubts!

Hey ! Thanks for the PR.

We need to make some changes in order to add this.

• First we should try to split the logic between the models. They behave differently, one returns texts and the other handles objects.
• For now let's keep it simple and let's keep the function generatePrompt returning a string for all models. So we can keep using the same UI for the prompt view.
• We should add a select input in the config panel (where you configure your API KEY) to be able to switch between models. This parameter should be saved and handled in the same way that the API KEY is being handled.
• For the functions handling the "chat/completions" endpoint, we shouldn't be doing any spliiting on the responses like we do for the completions, since the texts are already split into roles.

Let me know if you have any doubts!

Hi, I have pushed new changes regarding with your suggestions. Can you pls take a look on that?

[mangeloni30]

Hey !

I added you a review, let's change those for a switch in the other file.

Hey JC, I have pushed more changes, this time the structure is different. Let me know how does this look to you.

[mangeloni30]

Hi @apresmoi, I just added the following changes.

• "model" folder is inside of "openai" folder
• Rename textDavinci003Completions function to queryTextDavinci003Completions
• Rename gpt35TurboCompletions function to queryGpt35TurboCompletions
• Rename Dropdown file name to SelectInput
• Each query function (text-davinci-003 - gpt-3.5-turbo) is returning the chat object directly so in queryCompletionsChat function I have added a switch statement to return the result directly.

[apresmoi]

Hey thanks again for the PR and corrections. I tried it out and it seems to work as spected.

Merging this.