Rémi Louf
Rémi Louf
We would need to integrate the method directly on Hugging Face's servers.
A quick note, unrelated to the discussion. During internal evaluations, we've come across a couple of instances where the model "thinks" beyond the total context length. This makes me think...
There’s a conflict that needs to be resolved. We should be able to merge once it’s solved.
Im going to take a look now that there is no async loop running in outlines anymore.
> Are these examples not actually being tested, or what? Feel free to update the documentation so the examples are tested in CI, that would be a great contribution.
This issue should actually go in the `outlines-core` project. It could go in the README there.
I think that it makes sense for *generation* to set the default value to `False` (as long as it's documented). OpenAI forces you to set it to `False` manually, but...
"Unintuitive" **if** you know the JSON specification inside and out. We are not using the spec for what it was intended to do (parsing), so we can take small liberties...
Moving this issue to `outlines-core` as this is where schema validation is handled.