How to create Wav2Vec2 With Language model

This is cool!
Can you make the inference widget use the language model?