Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Fallback Model Support #516

Open
sydney-runkle opened this issue Dec 20, 2024 · 6 comments
Open

Fallback Model Support #516

sydney-runkle opened this issue Dec 20, 2024 · 6 comments
Assignees
@sydney-runkle
Labels
new models Support for new model(s)

Comments

@sydney-runkle
Copy link
Member

sydney-runkle commented Dec 20, 2024 •
edited
Loading

To mitigate issues with unpredictable LLM downtime: we'd like to add a new model that supports a list of models to use as fallbacks if the first doesn't work.
@sydney-runkle sydney-runkle added the new models Support for new model(s) label Dec 20, 2024
@samuelcolvin samuelcolvin added the good first issue Good for newcomers label Dec 20, 2024
The-CodeINN added a commit to The-CodeINN/pydantic-ai that referenced this issue Dec 23, 2024
@The-CodeINN
Add fallback models support
81b78ed 
Fixes pydantic#516

Add support for fallback models in the `Agent` class.

* **Agent Class Changes:**
  - Add `fallback_models` attribute to store a list of fallback models.
  - Modify `run`, `run_sync`, and `run_stream` methods to iterate through `fallback_models` if the primary model fails.
  - Update `_get_model` method to handle fallback models.

* **Model and AgentModel Classes Changes:**
  - Add `fallback_models` attribute to `Model` and `AgentModel` classes.

* **Tests:**
  - Add tests in `tests/test_agent.py` to verify the functionality of fallback models.
@The-CodeINN The-CodeINN mentioned this issue Dec 23, 2024
Closed
@samuelcolvin
Copy link
Member

samuelcolvin commented Jan 2, 2025 •
edited
Loading

I think the API can be as Simple as

model = FallbackModel('openai:gpt-4o', 'claude-3-5-haiku-latest', GeminiModel(...))
agent = Agent(model, ...)

E.g. the signature of FallbackModel is something like

class FallbackModel:
    def __init__(self, default_model: models.Model | models.KnownModelName, *fallback_models: models.Model | models.KnownModelName):
        ...

@samuelcolvin samuelcolvin removed the good first issue Good for newcomers label Jan 2, 2025
@phil65
Copy link

phil65 commented Jan 3, 2025

Implemented a pydantic-ai FallbackModel some time ago here, in case there is interest:
https://github.com/phil65/LLMling-models
Cheers!

@samuelcolvin
Copy link
Member

Another advantage of having a consistent way to detect 503 etc. responses, is we can use it to mark live tests as xfail when the underlying model fails.

@phil65
Copy link

phil65 commented Jan 7, 2025 •
edited
Loading

It´s actually quite funny what kind of things all can be done at model level. I now added a model to
to stream from another remote pydantic-ai model using FastAPI . 😄 (code might be sub-optimal though, just learning FastAPI a bit)

@sydney-runkle sydney-runkle self-assigned this Jan 8, 2025
@sydney-runkle
Copy link
Member Author

Will be starting work on this later today. PR to come soon 👍

@sydney-runkle sydney-runkle mentioned this issue Jan 8, 2025
Draft
@sydney-runkle
Copy link
Member Author

Implemented a pydantic-ai FallbackModel some time ago here, in case there is interest:
https://github.com/phil65/LLMling-models
Cheers!

Awesome, thanks so much @phil65!

@DaveOkpare DaveOkpare mentioned this issue Jan 10, 2025
Closed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@sydney-runkle sydney-runkle

Labels
new models Support for new model(s)
Projects
None yet
Milestone
No milestone
Development

Successfully merging a pull request may close this issue.

Add fallback models support The-CodeINN/pydantic-ai
3 participants
@phil65 @samuelcolvin @sydney-runkle