Thanks for this. We've been doing something similar with the Universal Sentence Encoder en masse (https://www.tensorflow.org/hub/tutorials/semantic_similarity_w

Thanks for this. We've been doing something similar with the Universal Sentence Encoder en masse (https://www.tensorflow.org/hub/tutorials/semantic_similarity_with_tf_hub_universal_encoder https://www.tensorflow.org/hub/tutorials/semantic_similarity...) Curious if anyone has recommendations on good stashes or datasets of already-encoded embeddings? This sounds geeky but to some extent, I dont even "care" about the original text but would love to just get the embedding vectors and play with those.