Inference with an external LM?
#69
by agonzalezd - opened
Greetings,
Is it possible to add an external Language Model (LM) during audio inference? It could be as simple as an n-gram or a LLM such as LLaMA or so.
Thanks in advance!
I think it is possible but is not implemented in the huggingface transformer generate function. You might have to implement yourself.