app.py

import streamlit as st
import numpy as np
from sentence_transformers import SentenceTransformer
from sklearn.metrics.pairwise import cosine_similarity
import spacy

left_text = st.text_area('First', 'This is a test')
right_text = st.text_area('Second', 'This is another test')

st.toast("Loading spacy...")
nlp = spacy.load("en_core_web_sm")

st.toast("Loading rufimelo/Legal-BERTimbau-sts-base...")
model = SentenceTransformer("rufimelo/Legal-BERTimbau-sts-base")

st.toast("Legal-BERTimbau-sts-base: computing embeddings...")
embeddings = model.encode([left_text, right_text])

st.toast("Legal-BERTimbau-sts-base: computing similarity...")
similarity = cosine_similarity(embeddings[: 1], embeddings[1 :])
st.info("Legal-BERTimbau-sts-base: score ->")
st.dataframe(similarity)

st.toast("Loading nlpaueb/legal-bert-base-uncased...")
model = SentenceTransformer("nlpaueb/legal-bert-base-uncased")

st.toast("legal-bert-base-uncased: computing embeddings...")
embeddings = model.encode([left_text, right_text])

st.toast("legal-bert-base-uncased: computing similarity...")
similarity = cosine_similarity(embeddings[: 1], embeddings[1 :])
st.info("legal-bert-base-uncased: score ->")
st.dataframe(similarity)