Add support for dependent types.

[notarealdeveloper]

Thanks to your help with #2, the gpts library now has a PR that adds a typed ask_for method using guidance and minml.

However, when me and @rskottap added tests to that method, we found it giving outputs that were never found in the (untyped) free response outputs.

Specifically, asking the models directly, without the guidance of guidance, we never got a negative value for the answer to any of the questions:

• "How many ounces are in a pound?"
• "How many days are in a week?"
• "How many eggs are in a dozen?"

Here's what we get with the minml based methods:

Any ideas on how to add a "positive" constraint to the int or float types, or more generally how to add other dependent types like "strings that match a regex"?

[hudson-ai]

Annotated types!

e.g.

from minml import gen_type
from typing import Annotated
from pydantic import StringConstraints

type = Annotated[str, StringConstraints(pattern=r'[A-Z]\d')]

Then use gen_type from there.

Note that

1. I only implemented string constraints -- it's left as an exercise to the reader if you want constraints on floats in the short-term. In the long-term, I may implement that myself.
2. guidance currently has very limited regex support, e.g. see the issue here: https://github.com/guidance-ai/guidance/issues/530

[notarealdeveloper]

Fantastic! We'll look into it and see what we can do.

[hudson-ai]

I find it somewhat concerning that you're seeing negative numbers in these contexts after adding guidance/minml to your workflow. Can you post a minimal example of a prompt that gives an erroneous negative? I would like to understand what's going on here :)