revierus-tech/LegalSimilarity

1import streamlit as st2import numpy as np3from sentence_transformers import SentenceTransformer4from sklearn.metrics.pairwise import cosine_similarity5import spacy6 7left_text = st.text_area('First', 'This is a test')8right_text = st.text_area('Second', 'This is another test')9 10st.toast("Loading spacy...")11nlp = spacy.load("en_core_web_sm")12 13st.toast("Loading rufimelo/Legal-BERTimbau-sts-base...")14model = SentenceTransformer("rufimelo/Legal-BERTimbau-sts-base")15 16st.toast("Legal-BERTimbau-sts-base: computing embeddings...")17embeddings = model.encode([left_text, right_text])18 19st.toast("Legal-BERTimbau-sts-base: computing similarity...")20similarity = cosine_similarity(embeddings[: 1], embeddings[1 :])21st.info("Legal-BERTimbau-sts-base: score ->")22st.dataframe(similarity)23 24st.toast("Loading nlpaueb/legal-bert-base-uncased...")25model = SentenceTransformer("nlpaueb/legal-bert-base-uncased")26 27st.toast("legal-bert-base-uncased: computing embeddings...")28embeddings = model.encode([left_text, right_text])29 30st.toast("legal-bert-base-uncased: computing similarity...")31similarity = cosine_similarity(embeddings[: 1], embeddings[1 :])32st.info("legal-bert-base-uncased: score ->")33st.dataframe(similarity)34