Update app.py

app.py

@@ -22,6 +22,9 @@ def fetch_dataset_and_init():
     df = string_counts_series.sort_values(ascending=False).to_frame()
     df.columns = ["count"]
     df = df.reset_index()
+    df = df.rename(columns={"index": "library_name"})
+
+    df.replace(to_replace=[None], value="No library_name", inplace=True)
     df_log = df.copy()
     df_log['count'] = np.log(df_log['count'])
 
@@ -37,24 +40,35 @@ def get_current_nb_models():
 
 plot_height = 512
 plot_width = 1512
+select_box = ["all"]
 top_k = len(df)
 
-def bar_plot_fn(display, top_k):
+def bar_plot_fn(display, top_k, select_box):
     if display == "simple":
+        if select_box is not None and ("all" not in select_box or select_box != ["all"]):
+            current_df = df[df["library_name"].isin(select_box)]
+        else:
+            current_df = df[:top_k]
+
         return gr.BarPlot(
-            df[:top_k],
-            x="index",
+            current_df,
+            x="library_name",
             y="count",
-            tooltip=["index", "count"],
+            tooltip=["library_name", "count"],
             height=plot_height,
             width=plot_width
         )
     elif display == "log":
+        if select_box is not None and ("all" not in select_box or select_box != ["all"]):
+            current_df = df_log[df_log["library_name"].isin(select_box)]
+        else:
+            current_df = df_log[:top_k]
+
         return gr.BarPlot(
-            df_log[:top_k],
-            x="index",
+            current_df,
+            x="library_name",
             y="count",
-            tooltip=["index", "count"],
+            tooltip=["library_name", "count"],
             height=plot_height,
             width=plot_width
         )
@@ -72,7 +86,7 @@ with gr.Blocks() as bar_plot:
                 label="Type of Bar Plot",
             )
             top_k = gr.Slider(
-                label="Select top-K most used library_name",
+                label="Select top-K most used library_name (This leads to a no-op if you selected something else than 'all' in the columns below)",
                 value=len(df),
                 minimum=1,
                 maximum=len(df),
@@ -85,8 +99,16 @@ with gr.Blocks() as bar_plot:
             fetch_button = gr.Button(value="Fetch current number of models without model cards (takes up to 1min to fetch everything)")
             text_box = gr.Textbox(value="", label="Number of models without model cards")
 
-    top_k.change(bar_plot_fn, inputs=[display, top_k], outputs=plot)
-    display.change(bar_plot_fn, inputs=[display, top_k], outputs=plot)
+        with gr.Column():
+            select_box = gr.Dropdown(
+                ["all"] + df["library_name"].tolist(), value=["all"], multiselect=True, label="Libraries to inspect", info="Select specific libraries to inspect"
+            )
+
+
+    top_k.change(bar_plot_fn, inputs=[display, top_k, select_box], outputs=plot)
+    display.change(bar_plot_fn, inputs=[display, top_k, select_box], outputs=plot)
+    select_box.change(bar_plot_fn, inputs=[display, top_k, select_box], outputs=plot)
+
     fetch_button.click(get_current_nb_models, outputs=[text_box])
     bar_plot.load(fn=bar_plot_fn, inputs=[display, top_k], outputs=plot)