bigscience-data/corpus-map

1import streamlit as st2import numpy as np3import plotly.express as px4import pandas as pd5st.set_page_config(page_title="BLOOM Corpus Treemap",layout='wide')6 7 8df = pd.read_pickle('bloom_dataset.pkl')9fig = px.treemap(df, path=[px.Constant("ROOTS"), 'Macroarea', 'Family', 'Genus', 'Language', 'dataset_name'],10                 values='num_bytes', maxdepth=4)11fig.update_traces(root_color="pink")12fig.update_layout(margin = dict(t=50, l=25, r=25, b=25))13 14st.plotly_chart(fig, use_container_width=True)15