Can guidance be used on custom models or NVIDIA NEMO model ?

[tcourat]

Let's say I have a custom decoder class in Pytorch, can I use this library with it ?

It seems to mainly interact with Hugginface Transformer framework and other common API (like openAI), but does not mention using a raw pytorch mordel.

I'm also interested in using guidance with NeMO for intent classification and slot filling : https://docs.nvidia.com/deeplearning/nemo/user-guide/docs/en/main/asr/speech_intent_slot/intro.html but it does not seem straightforward.