Restore grouped comparison chart

app.py

@@ -982,6 +982,36 @@ def metric_chart(results: pd.DataFrame, selected_metrics: list[str] | None = Non
     return fig
 
 
+def bar_chart(results: pd.DataFrame, selected_metrics: list[str] | None = None) -> go.Figure:
+    if results is None or results.empty:
+        return go.Figure()
+    selected_metrics = selected_metrics or METRIC_CHOICES
+    metric_cols = [c for c in selected_metrics if c in results.columns and results[c].notna().any()]
+    if not metric_cols:
+        metric_cols = [c for c in METRIC_CHOICES if c in results.columns and results[c].notna().any()]
+    if not metric_cols:
+        return go.Figure()
+    clean = results.sort_values("rank") if "rank" in results.columns else results.copy()
+    long = clean.melt(id_vars=["model"], value_vars=metric_cols, var_name="metric", value_name="score")
+    fig = px.bar(
+        long,
+        x="model",
+        y="score",
+        color="metric",
+        barmode="group",
+        color_discrete_sequence=GOOGLE_COLORS,
+        title="Grouped metric comparison",
+    )
+    fig.update_layout(
+        template="plotly_white",
+        height=360,
+        margin=dict(l=25, r=20, t=45, b=35),
+        legend_title_text="Metric",
+        hovermode="x unified",
+    )
+    return fig
+
+
 def time_chart(results: pd.DataFrame) -> go.Figure:
     if results is None or results.empty or "seconds" not in results:
         return go.Figure()
@@ -1020,7 +1050,7 @@ def run_catalog(
     df = get_dataset(dataset_name, sample_size, seed)
     tabfm_params = resolve_tabfm_params(tabfm_preset, tabfm_n_estimators, tabfm_max_rows, tabfm_max_features, tabfm_batch_size, tabfm_enable_nnls, tabfm_crosses, tabfm_svd, tabfm_max_eval_rows)
     results, preview, summary = benchmark_frame(df, spec.target, spec.task, sample_size, test_percent / 100, seed, selected_models, include_tabfm, tabfm_params)
-    return summary, results.round(4), metric_chart(results, selected_metrics, chart_style), time_chart(results), preview
+    return summary, results.round(4), metric_chart(results, selected_metrics, chart_style), time_chart(results), bar_chart(results, selected_metrics), preview
 
 
 def run_upload(
@@ -1050,7 +1080,7 @@ def run_upload(
     selected_task = None if task == "Auto" else task.lower()
     tabfm_params = resolve_tabfm_params(tabfm_preset, tabfm_n_estimators, tabfm_max_rows, tabfm_max_features, tabfm_batch_size, tabfm_enable_nnls, tabfm_crosses, tabfm_svd, tabfm_max_eval_rows)
     results, preview, summary = benchmark_frame(df, target, selected_task, sample_size, test_percent / 100, seed, selected_models, include_tabfm, tabfm_params)
-    return summary, results.round(4), metric_chart(results, selected_metrics, chart_style), time_chart(results), preview
+    return summary, results.round(4), metric_chart(results, selected_metrics, chart_style), time_chart(results), bar_chart(results, selected_metrics), preview
 
 
 def redraw_metric_chart(results: pd.DataFrame, selected_metrics: list[str], chart_style: str):
@@ -1059,6 +1089,12 @@ def redraw_metric_chart(results: pd.DataFrame, selected_metrics: list[str], char
     return metric_chart(pd.DataFrame(results), selected_metrics, chart_style)
 
 
+def redraw_bar_chart(results: pd.DataFrame, selected_metrics: list[str]):
+    if results is None or len(results) == 0:
+        return go.Figure()
+    return bar_chart(pd.DataFrame(results), selected_metrics)
+
+
 def catalog_table() -> pd.DataFrame:
     return pd.DataFrame(
         [
@@ -1136,6 +1172,7 @@ def build_app() -> gr.Blocks:
                             with gr.Row():
                                 chart = gr.Plot(label="Metric comparison")
                                 speed = gr.Plot(label="Speed")
+                            bars = gr.Plot(label="Grouped comparison")
                             preview = gr.Dataframe(label="Held-out preview", interactive=False)
                     run_inputs = [
                         dataset,
@@ -1156,10 +1193,12 @@ def build_app() -> gr.Blocks:
                         tabfm_svd,
                         tabfm_max_eval_rows,
                     ]
-                    run_btn.click(run_catalog, run_inputs, [summary, leaderboard, chart, speed, preview])
+                    run_outputs = [summary, leaderboard, chart, speed, bars, preview]
+                    run_btn.click(run_catalog, run_inputs, run_outputs)
                     metric_toggles.change(redraw_metric_chart, [leaderboard, metric_toggles, chart_style], chart)
+                    metric_toggles.change(redraw_bar_chart, [leaderboard, metric_toggles], bars)
                     chart_style.change(redraw_metric_chart, [leaderboard, metric_toggles, chart_style], chart)
-                    demo.load(run_catalog, run_inputs, [summary, leaderboard, chart, speed, preview])
+                    demo.load(run_catalog, run_inputs, run_outputs)
                 with gr.Tab("Upload Dataset"):
                     with gr.Row():
                         with gr.Column(scale=1, elem_classes=["panel"]):
@@ -1190,6 +1229,7 @@ def build_app() -> gr.Blocks:
                             with gr.Row():
                                 upload_chart = gr.Plot(label="Metric comparison")
                                 upload_speed = gr.Plot(label="Speed")
+                            upload_bars = gr.Plot(label="Grouped comparison")
                             upload_preview = gr.Dataframe(label="Held-out preview", interactive=False)
                     upload_btn.click(
                         run_upload,
@@ -1214,9 +1254,10 @@ def build_app() -> gr.Blocks:
                             upload_tabfm_svd,
                             upload_tabfm_max_eval_rows,
                         ],
-                        [upload_summary, upload_leaderboard, upload_chart, upload_speed, upload_preview],
+                        [upload_summary, upload_leaderboard, upload_chart, upload_speed, upload_bars, upload_preview],
                     )
                     upload_metric_toggles.change(redraw_metric_chart, [upload_leaderboard, upload_metric_toggles, upload_chart_style], upload_chart)
+                    upload_metric_toggles.change(redraw_bar_chart, [upload_leaderboard, upload_metric_toggles], upload_bars)
                     upload_chart_style.change(redraw_metric_chart, [upload_leaderboard, upload_metric_toggles, upload_chart_style], upload_chart)
                 with gr.Tab("Dataset Catalog"):
                     gr.Dataframe(catalog_table(), interactive=False, label="Included benchmark catalog")