Add checkbox in Results to hide stderr

app.py

@@ -63,6 +63,7 @@ with gr.Blocks(fill_height=True, fill_width=True) as demo:
                 lines=3,
                 visible=False,
             )
+            hide_errors = gr.Checkbox(label="Hide Errors", value=True, info="Options")
             results = gr.HTML()
         with gr.Tab("Configs"):
             load_configs_btn = gr.Button("Load", interactive=False)
@@ -140,9 +141,9 @@ with gr.Blocks(fill_height=True, fill_width=True) as demo:
     )
     # Display results
     gr.on(
-        triggers=[dataframe_1.change, dataframe_2.change, results_task.change],
+        triggers=[dataframe_1.change, dataframe_2.change, results_task.change, hide_errors.change],
         fn=display_results,
-        inputs=[results_task, dataframe_1, dataframe_2],
+        inputs=[results_task, hide_errors, dataframe_1, dataframe_2],
         outputs=[results, configs],
     )
     gr.on(

src/results.py

@@ -51,16 +51,19 @@ async def load_results_dataframes(*model_ids, result_paths_per_model=None):
     return result
 
 
-def display_results(task, *dfs):
+def display_results(task, hide_errors, *dfs):
     dfs = [df.set_index("index") for df in dfs if "index" in df.columns]
     if not dfs:
         return None, None
     df = pd.concat(dfs)
     df = df.T.rename_axis(columns=None)
-    return display_tab("results", df, task), display_tab("configs", df, task)
+    return (
+        display_tab("results", df, task, hide_errors=hide_errors),
+        display_tab("configs", df, task),
+    )
 
 
-def display_tab(tab, df, task):
+def display_tab(tab, df, task, hide_errors=True):
     df = df.style.format(escape="html", na_rep="")
     df.hide(
         [
@@ -75,6 +78,8 @@ def display_tab(tab, df, task):
                     if task != "All"
                     else row.startswith(f"{tab}.leaderboard_arc_challenge")
                 )
+                # Hide errors
+                or (hide_errors and row.endswith("_stderr,none"))
             )
         ],
         axis="index",