Text WebUI

[holdengand]

I'm starting to feel like an idiot.

Langflow is able to be used with oobabooga right? Considering how langflow is quite popular when it comes to these kinds of frameworks; yet hardly anyone has talked about using it with Text WebUI. The few conversations I have found were nearly a year ago; and were either inconclusive or seemed to imply it was possible but not implemented yet officially.

Its been around a year, so... how? I got oobabooga running with the API, along with a api key. But if I use the OpenAI component it seems to just want to use one of ChatGPT's models. So... can this be done? Does it need a custom component? Am I missing something obvious? None of the other components seem remotely related.

I have used langflow with ollama in the past; but was hoping to mess with it again using something more flexible now.

[dosubot[bot]]

To integrate Langflow with oobabooga or similar APIs, you would likely need to create a custom component. Here's a general process you can follow to achieve this:

1. Create a Custom Component: Define a new class, such as OobaboogaAPIComponent, that inherits from the Component class in Langflow.

2. Define Component Metadata: Include attributes like display_name, description, and icon to describe your component.

3. Specify Inputs and Outputs: Use the inputs and outputs lists to define the necessary inputs (e.g., API key, endpoint URL) and outputs for your component.

4. Implement Processing Logic: Write the logic to interact with the oobabooga API. This might involve sending requests to the API and processing the responses.

5. Add the Component to Langflow: Place your component code in the appropriate directory, such as /components/apis, and update the __init__.py file to include your new component.

6. Update Dependencies: If your component requires additional Python packages, add them to the pyproject.toml file.

7. Test and Document: Test your component to ensure it works as expected and provide documentation to help others use it.

This approach allows you to customize Langflow to work with APIs beyond those officially supported, like OpenAI. If you have previously used Langflow with other models like ollama, you might find this process familiar.

To continue talking to Dosu, mention @dosu.

[holdengand]

So no?
No one uses Text Gen WebUI with LangFlow I guess.

[anovazzi1]

hey, you have some options:
1 create a custom component for your usecase
2 if you have a WebUI api provider that follows the openAi schema you can use the openAi component and replace the openAI base with the endpoint of your api
3 you can create a real component in the components folder for the WebUI and open a pr so we can add it on the next release :)