sentence-transformers How to use Culture-and-Morality-Lab/psyembedding-roberta-large with sentence-transformers:
from sentence_transformers import SentenceTransformer
model = SentenceTransformer("Culture-and-Morality-Lab/psyembedding-roberta-large")
sentences = [
"So when are they leaving? I saw the protest had smaller amounts of protestors today as opposed to Friday and Saturday",
"Hillary's great, but center-leftists seem to do better with younger candidates. Bill, Blair, Obama, Trudeau, and now Macron. She'll be 72 in 2020.",
". I felt very sorry for you during your meltdown on He drove you insane but, of course, Piers is a lot smarter than you",
"As a kid, my friends and I all believed that Gymkata was the most violent, bloody movie ever made. I'm not sure who started that rumor. It was probably born out of the frustration of 10 year olds who weren't allowed to see it for one reason or other. Years after Gymkata was released, it became a perennial late night cable movie, and as a result, I've been able to make up for lost time. I must have seen scenes from this dreadful excuse for a film over a dozen times, and I can always spot it from 1-2 seconds of screen time. However, aside from the forced coupling of gymnastics and martial arts, the bad dubbing, the stiff dialog, and the outrageously difficult story-line, the film has some things going for it. With all that's bad about the movie visually, the sound is actually pretty entertaining. Never before has a punch or kick landed with so little force and so much volume! The canned kung-fu sounds are cheeky, but the slowed and pitched-down music, and the nearly 5 minute slow motion scene are truly weird. The chase through the city of demented, blood-thirsty villagers isn't really tense as much as it is irritating, and there are enough bad wigs and extras who all but look into the camera and wave to make this train-wreck a little fun. Could it be headed for cult-classic status? Where is MST3K when we need it?"
]
embeddings = model.encode(sentences)
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [4, 4] 
