app.py

import streamlit as st

from joblib import load
import pandas as pd
import plotly.express as px


reduced_embeddings = load("embeddings_tsne.joblib")
combined_sentences = load("sentence_list.joblib")
languages = load("language_list.joblib")
clusters = load("clusters.joblib")

def main():
    st.title("LASER Multilingual Sentence Embeddings Visualization")

    df = pd.DataFrame({
        'TSNE Component 1': reduced_embeddings[:, 0],
        'TSNE Component 2': reduced_embeddings[:, 1],
        'Language': languages,
        'Sentence': combined_sentences,
        'Cluster': ['Cluster {}'.format(cluster) for cluster in clusters]
    })

    select_all = st.checkbox("Select All Clusters")

    unique_clusters = df['Cluster'].unique()

    # Active only if 'Select All Clusters' is unchecked
    if not select_all:
        selected_clusters = st.multiselect("Select clusters to display", unique_clusters, default=unique_clusters[:10])
    else:
        selected_clusters = unique_clusters

    filtered_df = df[df['Cluster'].isin(selected_clusters)]

    fig = px.scatter(filtered_df, x='TSNE Component 1', y='TSNE Component 2', 
                     color='Language', hover_data=['Sentence', 'Cluster'])
    fig.update_layout(title="Multilingual Sentence Embeddings Visualization",
                      xaxis_title="TSNE Component 1", yaxis_title="TSNE Component 2",
                      legend_title="Language")

    st.plotly_chart(fig, use_container_width=True)

if __name__ == "__main__":
    main()