I'm trying to freeze the Quick-thoughts model but am having trouble identifying the input and output nodes. I believe the output node should be "word_embeddings" but am not 100% sure.
My end goal is to be able to use the encoder from a frozen version of the model.