Add a comparison between LMQL and Guidance

[jvhoffbauer]

I would like to understand how Guidance compares to LMQL when reading the docs. Can we gather a list of additional features?

[slundberg]

People should feel free to leave notes here. The core of Guidance was written before LMQL (but not widely released), so the design choices were made independently and I expect there is stuff to learn beyond just the fundamental difference of a template language style vs. a SQL database style.

[jvhoffbauer]

Makes sense! Happy to start creating a list.

Do I understand correctly that LMQL especially does not allow pattern enforcing in inference?

[marcotcr]

We are not keeping track of LMQL, so we don't know what they do or do not support :)

[charles-dyfis-net]

@jvhoffbauer, my understanding is that LMQL can enforce any constraint that can be represented as a context-free grammar.

[ogencoglu]

We are not keeping track of LMQL, so we don't know what they do or do not support :)

Of course you do! Let's don't kid ourselves. Anyway, here is a comparison by LMQL: https://docs.lmql.ai/en/latest/python/comparison.html

[ianarawjo]

I was also curious about this. Given that guidance came later, it appeared to me, and other people as well, as a kind of knock-off of LMQL, except w a big corporation behind it. Maybe this perception is wrong —but, still, it would be nice to have a comparison somewhere to LMQL and pay respect where respect is due to the initial originators of the idea.

[ogencoglu]

LMQL is more open and less dismissive about such comparison and they provided a documentation related to that: https://docs.lmql.ai/en/latest/python/comparison.html