Samuel Colvin

I've tried running PydanticAI with Ollama, and it's awesome. Thanks @JojoJr24, I took your code and tweaked it slightly. Really looking forward to a nice wrapper for running this without...

happy to add another example.

@benomahony what are you talking about? https://github.com/pydantic/pydantic-ai/blob/0a5c40d3b5014245c95ea8f043566a0bbe1b48b2/pydantic_ai_slim/pydantic_ai/models/openai.py#L80

we need to document the behaviour, and add a warning when the parents are used in the "wrong" order.

Thanks, but I don't think the underlying rust library allows this. If I'm wrong about that, I'd be happy to look into this or accept a PR. Otherwise three suggestions:...

Humm, I think the behaviour we have now is fine. I think it’s perfectly reasonable for `1.` to be interpreted as an invalid value and dropped (sp `[1.1, 2.2, 3.`...

or presumably `123e` will be dropped from partial parsing even though it could be the beginning of a value number.

I think the correct thing to do is drop any incomplete number. The thing is that `10` is not similar to or a good proxy for `1024` in the way...

Sorry for the lack of reply on this. We've discussed it at length internally, and it's a priority for us. We'll get back to you later today with more details...

Hi, we'll need to see the model you're using, and an example from of the input data to be able to help. Also, how are you instrumenting the endpoint and...