Spaces:

bigscience-data
/

corpus-map

Running

christopher commited on May 31, 2022

Commit

2dac9f9

1 Parent(s): a8bb596

Update app.py

Files changed (1) hide show

app.py CHANGED Viewed

@@ -6,7 +6,7 @@ st.set_page_config(page_title="BLOOM Corpus Treemap",layout='wide')
 df = pd.read_pickle('bloom_dataset.pkl')
-df['normalized_size'] = np.sqrt(df['num_bytes'])
 fig = px.treemap(df, path=[px.Constant("Bloom Corpus"), 'Macroarea', 'Family', 'Genus', 'Language', 'dataset_name'],
                  values='normalized_size', maxdepth=3)
 fig.update_traces(root_color="pink")

 df = pd.read_pickle('bloom_dataset.pkl')
+df['normalized_size'] = np.log(df['num_bytes'])
 fig = px.treemap(df, path=[px.Constant("Bloom Corpus"), 'Macroarea', 'Family', 'Genus', 'Language', 'dataset_name'],
                  values='normalized_size', maxdepth=3)
 fig.update_traces(root_color="pink")