Improve model controls and TabFM visibility

README.md

@@ -42,7 +42,9 @@ The built-in catalog uses real open datasets where possible:
 
 TabFM is loaded through the public Google Research package when available. The app keeps graceful fallbacks so the Space still works on CPU-only or dependency-constrained runtimes. Live TabFM runs are opt-in because the model weights are large and CPU-only inference can be slow.
 
-Model choices include Logistic/Ridge, RandomForest, ExtraTrees, GradientBoosting, HistGradientBoosting, AdaBoost, KNN, SVM, XGBoost, LightGBM, Dummy, and optional live TabFM.
+Model choices include Logistic, Linear Regression, Ridge, Bayesian Ridge, RandomForest, ExtraTrees, GradientBoosting, HistGradientBoosting, AdaBoost, KNN, SVC/SVR, XGBoost, LightGBM, Dummy, and optional live TabFM.
+
+TabFM is always represented in the leaderboard: it appears as skipped until **Run TabFM live** is enabled, then reports live metrics when the run completes.
 
 Note: TabFM's weights use their own non-commercial license. Review the upstream model license before using this Space commercially.
 

app.py

@@ -31,7 +31,7 @@ from sklearn.ensemble import (
     RandomForestRegressor,
 )
 from sklearn.impute import SimpleImputer
-from sklearn.linear_model import LogisticRegression, Ridge
+from sklearn.linear_model import BayesianRidge, LinearRegression, LogisticRegression, Ridge
 from sklearn.metrics import (
     accuracy_score,
     f1_score,
@@ -734,19 +734,21 @@ def available_baselines(task: str) -> dict[str, object]:
             "HistGradientBoosting": Pipeline([("prep", make_preprocessor(pd.DataFrame())), ("model", HistGradientBoostingClassifier(max_iter=120, random_state=RANDOM_STATE))]),
             "AdaBoost": Pipeline([("prep", make_preprocessor(pd.DataFrame())), ("model", AdaBoostClassifier(n_estimators=80, learning_rate=0.08, random_state=RANDOM_STATE))]),
             "KNN": Pipeline([("prep", make_preprocessor(pd.DataFrame(), True)), ("model", KNeighborsClassifier(n_neighbors=7))]),
-            "SVM": Pipeline([("prep", make_preprocessor(pd.DataFrame(), True)), ("model", SVC(C=1.0, probability=True, random_state=RANDOM_STATE))]),
+            "SVC": Pipeline([("prep", make_preprocessor(pd.DataFrame(), True)), ("model", SVC(C=1.0, probability=True, random_state=RANDOM_STATE))]),
             "Dummy": Pipeline([("prep", make_preprocessor(pd.DataFrame())), ("model", DummyClassifier(strategy="most_frequent"))]),
         }
     else:
         models = {
+            "LinearRegression": Pipeline([("prep", make_preprocessor(pd.DataFrame(), True)), ("model", LinearRegression())]),
             "Ridge": Pipeline([("prep", make_preprocessor(pd.DataFrame(), True)), ("model", Ridge(alpha=1.0))]),
+            "BayesianRidge": Pipeline([("prep", make_preprocessor(pd.DataFrame(), True)), ("model", BayesianRidge())]),
             "RandomForest": Pipeline([("prep", make_preprocessor(pd.DataFrame())), ("model", RandomForestRegressor(n_estimators=80, min_samples_leaf=2, n_jobs=-1, random_state=RANDOM_STATE))]),
             "ExtraTrees": Pipeline([("prep", make_preprocessor(pd.DataFrame())), ("model", ExtraTreesRegressor(n_estimators=120, min_samples_leaf=2, n_jobs=-1, random_state=RANDOM_STATE))]),
             "GradientBoosting": Pipeline([("prep", make_preprocessor(pd.DataFrame())), ("model", GradientBoostingRegressor(n_estimators=100, learning_rate=0.06, random_state=RANDOM_STATE))]),
             "HistGradientBoosting": Pipeline([("prep", make_preprocessor(pd.DataFrame())), ("model", HistGradientBoostingRegressor(max_iter=120, random_state=RANDOM_STATE))]),
             "AdaBoost": Pipeline([("prep", make_preprocessor(pd.DataFrame())), ("model", AdaBoostRegressor(n_estimators=80, learning_rate=0.08, random_state=RANDOM_STATE))]),
             "KNN": Pipeline([("prep", make_preprocessor(pd.DataFrame(), True)), ("model", KNeighborsRegressor(n_neighbors=7))]),
-            "SVM": Pipeline([("prep", make_preprocessor(pd.DataFrame(), True)), ("model", SVR(C=1.0))]),
+            "SVR": Pipeline([("prep", make_preprocessor(pd.DataFrame(), True)), ("model", SVR(C=1.0))]),
             "Dummy": Pipeline([("prep", make_preprocessor(pd.DataFrame())), ("model", DummyRegressor(strategy="median"))]),
         }
     if importlib.util.find_spec("xgboost"):
@@ -889,6 +891,7 @@ def benchmark_frame(
 
     rows: list[dict[str, object]] = []
     notes: list[str] = []
+    selected_models = selected_models or []
     models = available_baselines(task)
     for name, model in models.items():
         if name not in selected_models:
@@ -905,7 +908,7 @@ def benchmark_frame(
             rows.append({"model": name, "status": f"failed: {exc}", "seconds": time.perf_counter() - start})
     for name in selected_models:
         if name not in models:
-            rows.append({"model": name, "status": "unavailable in this runtime", "seconds": 0.0, "rank_score": np.nan})
+            rows.append({"model": name, "status": f"not compatible with {task} or unavailable", "seconds": 0.0, "rank_score": np.nan})
 
     if include_tabfm:
         start = time.perf_counter()
@@ -931,6 +934,7 @@ def benchmark_frame(
             notes.append(f"TabFM did not run in this environment. On Spaces, keep Python 3.11 and allow the GitHub dependency plus model download for `{TABFM_MODEL_ID}`.")
     else:
         rows.append({"model": "TabFM", "status": "skipped - enable Run TabFM live", "seconds": 0.0, "rank_score": np.nan})
+        notes.append("TabFM is listed as skipped because **Run TabFM live** is off. Enable it to benchmark TabFM; the first run may download large model weights.")
 
     results = pd.DataFrame(rows)
     metric_cols = [c for c in ["accuracy", "f1_weighted", "roc_auc", "rmse", "mae", "r2", "rank_score", "seconds"] if c in results.columns]
@@ -1173,14 +1177,17 @@ def catalog_table() -> pd.DataFrame:
 
 DEFAULT_MODELS = [
     "Logistic",
+    "LinearRegression",
     "Ridge",
+    "BayesianRidge",
     "RandomForest",
     "ExtraTrees",
     "GradientBoosting",
     "HistGradientBoosting",
     "AdaBoost",
     "KNN",
-    "SVM",
+    "SVC",
+    "SVR",
     "XGBoost",
     "LightGBM",
     "Dummy",
@@ -1200,7 +1207,14 @@ def build_app() -> gr.Blocks:
     .stat-card .value { font-size: 28px; font-weight: 760; margin-top: 12px; }
     .stat-card .trend { display: inline-block; margin-left: 8px; font-size: 12px; color: #027a48; background: #ecfdf3; border-radius: 999px; padding: 2px 8px; }
     .panel { background: #fff; border: 1px solid #edf0f5; border-radius: 14px; padding: 14px; box-shadow: 0 12px 35px rgba(15, 23, 42, 0.04); }
-    .gr-button-primary { background: #f97316 !important; border-color: #f97316 !important; }
+    .control-panel { background: linear-gradient(180deg, #ffffff 0%, #fbfcff 100%); border-top: 4px solid #f97316; }
+    .control-panel label span { color: #344054; font-weight: 720; }
+    .control-panel input, .control-panel textarea, .control-panel select { border-radius: 10px !important; }
+    .control-panel .wrap { gap: 9px !important; }
+    .control-panel .token, .control-panel [data-testid="token"] { background: #fff7ed !important; color: #c2410c !important; border: 1px solid #fed7aa !important; border-radius: 999px !important; }
+    .control-panel .checkbox label { border-radius: 999px !important; }
+    .gr-button-primary { background: linear-gradient(135deg, #f97316, #ea4335) !important; border-color: #f97316 !important; box-shadow: 0 12px 24px rgba(249, 115, 22, 0.25) !important; border-radius: 12px !important; min-height: 46px !important; font-weight: 760 !important; }
+    .plot-container, .table-wrap { border-radius: 14px !important; overflow: hidden; }
     footer { display: none !important; }
     """
     with gr.Blocks(title=APP_TITLE, css=css, theme=gr.themes.Soft(primary_hue="orange", secondary_hue="violet")) as demo:
@@ -1220,14 +1234,14 @@ def build_app() -> gr.Blocks:
             with gr.Tabs():
                 with gr.Tab("Arena"):
                     with gr.Row():
-                        with gr.Column(scale=1, elem_classes=["panel"]):
+                        with gr.Column(scale=1, elem_classes=["panel", "control-panel"]):
                             dataset = gr.Dropdown(dataset_names(), value="Titanic Survival", label="Dataset")
                             sample = gr.Slider(100, 50000, value=1000, step=100, label="Sample size")
                             test_pct = gr.Slider(10, 40, value=25, step=5, label="Test split (%)")
                             seed = gr.Number(value=42, precision=0, label="Random seed")
-                            models = gr.CheckboxGroup(DEFAULT_MODELS, value=DEFAULT_SELECTED_MODELS, label="Baselines")
+                            models = gr.Dropdown(DEFAULT_MODELS, value=DEFAULT_SELECTED_MODELS, multiselect=True, label="Models")
                             include_tabfm = gr.Checkbox(value=False, label="Run TabFM live (adds TabFM row)")
-                            metric_toggles = gr.CheckboxGroup(METRIC_CHOICES, value=["rmse", "mae", "r2", "accuracy", "f1_weighted", "roc_auc"], label="Chart metrics")
+                            metric_toggles = gr.Dropdown(METRIC_CHOICES, value=["rmse", "mae", "r2", "accuracy", "f1_weighted", "roc_auc"], multiselect=True, label="Chart metrics")
                             chart_style = gr.Radio(["Line", "Radar"], value="Line", label="Chart style")
                             with gr.Accordion("TabFM tuning", open=False):
                                 tabfm_preset = gr.Dropdown(list(TABFM_PRESETS.keys()) + ["Custom"], value="Fast", label="Preset")
@@ -1275,16 +1289,16 @@ def build_app() -> gr.Blocks:
                     demo.load(run_catalog, run_inputs, run_outputs)
                 with gr.Tab("Upload Dataset"):
                     with gr.Row():
-                        with gr.Column(scale=1, elem_classes=["panel"]):
+                        with gr.Column(scale=1, elem_classes=["panel", "control-panel"]):
                             file = gr.File(label="CSV file", file_types=[".csv"])
                             target = gr.Textbox(label="Target column")
                             task = gr.Radio(["Auto", "Classification", "Regression"], value="Auto", label="Task")
                             upload_sample = gr.Slider(100, 50000, value=1000, step=100, label="Sample size")
                             upload_test_pct = gr.Slider(10, 40, value=25, step=5, label="Test split (%)")
                             upload_seed = gr.Number(value=42, precision=0, label="Random seed")
-                            upload_models = gr.CheckboxGroup(DEFAULT_MODELS, value=DEFAULT_SELECTED_MODELS, label="Baselines")
+                            upload_models = gr.Dropdown(DEFAULT_MODELS, value=DEFAULT_SELECTED_MODELS, multiselect=True, label="Models")
                             upload_tabfm = gr.Checkbox(value=False, label="Run TabFM live (adds TabFM row)")
-                            upload_metric_toggles = gr.CheckboxGroup(METRIC_CHOICES, value=["rmse", "mae", "r2", "accuracy", "f1_weighted", "roc_auc"], label="Chart metrics")
+                            upload_metric_toggles = gr.Dropdown(METRIC_CHOICES, value=["rmse", "mae", "r2", "accuracy", "f1_weighted", "roc_auc"], multiselect=True, label="Chart metrics")
                             upload_chart_style = gr.Radio(["Line", "Radar"], value="Line", label="Chart style")
                             with gr.Accordion("TabFM tuning", open=False):
                                 upload_tabfm_preset = gr.Dropdown(list(TABFM_PRESETS.keys()) + ["Custom"], value="Fast", label="Preset")