added exports, fixes to search with a split and other qol

README.md

@@ -1,12 +1,13 @@
 ---
-title: Eee Test
+title: Every Eval Ever Space
 emoji: 👀
+python-version: 3.13
 colorFrom: pink
 colorTo: purple
 sdk: gradio
-sdk_version: 5.49.1
+sdk_version: 6.1.0
 app_file: app.py
-pinned: false
+pinned: true
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

app.py

@@ -131,6 +131,72 @@ def get_model_suggestions(query):
     return gr.update(choices=matches[:15])
 
 
+def export_leaderboard_to_csv(full_df, selected_leaderboard, search_query, selected_columns):
+    """Export the current leaderboard view to CSV."""
+    if full_df.empty:
+        return None
+    
+    df = full_df.copy()
+    
+    # Apply column selection
+    if selected_columns:
+        cols = ["Model"] + [c for c in df.columns if c in selected_columns and c != "Model"]
+        df = df[cols]
+    
+    # Apply search filter
+    if search_query:
+        mask = df.astype(str).apply(lambda row: row.str.contains(search_query, case=False, na=False).any(), axis=1)
+        df = df[mask]
+    
+    # Save to CSV with absolute path
+    from pathlib import Path
+    import tempfile
+    temp_dir = Path(tempfile.gettempdir())
+    filename = temp_dir / f"{selected_leaderboard.replace(' ', '_')}_leaderboard.csv"
+    df.to_csv(filename, index=False)
+    return str(filename)
+
+
+def export_comparison_to_csv(selected_models):
+    """Export model comparison to CSV."""
+    if not selected_models:
+        return None
+    
+    all_results = {}
+    for model_name in selected_models:
+        results, _ = search_model_across_leaderboards(model_name)
+        if results:
+            matched_model = list(results.keys())[0]
+            all_results[matched_model] = results[matched_model]
+    
+    if not all_results:
+        return None
+    
+    # Build comparison table
+    rows = []
+    for model_name, model_data in all_results.items():
+        for leaderboard_name, data in model_data.items():
+            results = data.get("results", {})
+            row = {
+                "Model": model_name,
+                "Leaderboard": leaderboard_name,
+                "Developer": data.get("developer"),
+                "Params (B)": data.get("params"),
+                "Architecture": data.get("architecture"),
+                "Precision": data.get("precision")
+            }
+            row.update(results)
+            rows.append(row)
+    
+    df = pd.DataFrame(rows)
+    from pathlib import Path
+    import tempfile
+    temp_dir = Path(tempfile.gettempdir())
+    filename = temp_dir / "model_comparison.csv"
+    df.to_csv(filename, index=False)
+    return str(filename)
+
+
 load_hf_dataset_on_startup()
 
 initial_leaderboards = get_available_leaderboards()
@@ -179,6 +245,8 @@ with gr.Blocks(title="Every Eval Ever", theme=get_theme(), css=get_custom_css())
                         )
                     with gr.Column(scale=1, min_width=120):
                         refresh_btn = gr.Button("↻ Refresh", variant="secondary", size="sm")
+                    with gr.Column(scale=1, min_width=120):
+                        export_btn = gr.DownloadButton("📥 Export CSV", variant="secondary", size="sm")
 
                 search_box = gr.Textbox(
                     label="Filter",
@@ -225,15 +293,19 @@ with gr.Blocks(title="Every Eval Ever", theme=get_theme(), css=get_custom_css())
                 </div>
             """
             
+            model_search_box = gr.Textbox(
+                label="Type to search for models",
+                placeholder="Start typing model name (e.g., gpt, llama, claude)...",
+                interactive=True,
+            )
+            
             with gr.Row(elem_classes="controls-bar"):
                 with gr.Column(scale=4):
-                    all_models = get_all_model_names()
                     model_dropdown = gr.Dropdown(
-                        choices=all_models,
-                        label="Search models to add",
+                        choices=[],
+                        label="Select from search results",
                         interactive=True,
                         allow_custom_value=False,
-                        filterable=True,
                     )
                 with gr.Column(scale=1, min_width=100):
                     clear_models_btn = gr.Button("Clear All", variant="secondary", size="sm")
@@ -246,6 +318,12 @@ with gr.Blocks(title="Every Eval Ever", theme=get_theme(), css=get_custom_css())
                 elem_classes="selected-models-group"
             )
             
+            with gr.Row():
+                with gr.Column(scale=4):
+                    pass
+                with gr.Column(scale=1, min_width=120):
+                    export_comparison_btn = gr.DownloadButton("📥 Export CSV", variant="secondary", size="sm")
+            
             radar_view = gr.Plot(label="Radar Comparison")
             model_card_view = gr.HTML(value=default_compare_html)
     
@@ -329,12 +407,19 @@ Submit via GitHub Pull Request:
         outputs=[leaderboard_selector]
     )
     
+    export_btn.click(
+        fn=export_leaderboard_to_csv,
+        inputs=[full_df_state, leaderboard_selector, search_box, column_selector],
+        outputs=[export_btn]
+    )
+    
     def add_model_and_compare(selected_model, current_selected):
         if not selected_model:
             comparison_html, plot = compare_models(current_selected) if current_selected else (default_compare_html, None)
             return (
                 current_selected,
-                gr.update(value=None),
+                gr.update(value=""),
+                gr.update(value=None, choices=[]),
                 gr.update(choices=current_selected, value=current_selected),
                 comparison_html,
                 plot
@@ -347,7 +432,8 @@ Submit via GitHub Pull Request:
         
         return (
             current_selected,
-            gr.update(value=None),
+            gr.update(value=""),
+            gr.update(value=None, choices=[]),
             gr.update(choices=current_selected, value=current_selected),
             comparison_html,
             plot
@@ -360,16 +446,23 @@ Submit via GitHub Pull Request:
     def clear_all_models():
         return (
             [],
-            gr.update(value=None),
+            gr.update(value=""),
+            gr.update(value=None, choices=[]),
             gr.update(choices=[], value=[]),
             default_compare_html,
             None
         )
     
+    model_search_box.input(
+        fn=get_model_suggestions,
+        inputs=[model_search_box],
+        outputs=[model_dropdown]
+    )
+    
     model_dropdown.select(
         fn=add_model_and_compare,
         inputs=[model_dropdown, selected_models_state],
-        outputs=[selected_models_state, model_dropdown, selected_models_group, model_card_view, radar_view]
+        outputs=[selected_models_state, model_search_box, model_dropdown, selected_models_group, model_card_view, radar_view]
     )
     
     selected_models_group.change(
@@ -380,7 +473,13 @@ Submit via GitHub Pull Request:
     
     clear_models_btn.click(
         fn=clear_all_models,
-        outputs=[selected_models_state, model_dropdown, selected_models_group, model_card_view, radar_view]
+        outputs=[selected_models_state, model_search_box, model_dropdown, selected_models_group, model_card_view, radar_view]
+    )
+    
+    export_comparison_btn.click(
+        fn=export_comparison_to_csv,
+        inputs=[selected_models_state],
+        outputs=[export_comparison_btn]
     )
     
     DATA_DIR.mkdir(exist_ok=True)

ui_components.py

@@ -177,7 +177,7 @@ body, .gradio-container {
 
 .metrics-section {
     margin-top: 1.25rem;
-    padding-top: 1.25rem;
+    padding: 1.25rem 1rem;
     border-top: 1px solid #e5e5e5;
 }
 
@@ -196,6 +196,29 @@ body, .gradio-container {
     gap: 0.75rem;
 }
 
+@media (max-width: 768px) {
+    .metrics-grid {
+        grid-template-columns: repeat(auto-fill, minmax(160px, 1fr));
+        gap: 0.5rem;
+    }
+    
+    .metric-card-header {
+        padding: 0.65rem 0.8rem;
+        flex-direction: column;
+        align-items: flex-start;
+        gap: 0.25rem;
+    }
+    
+    .metric-card-body {
+        padding: 0.65rem 0.8rem;
+        font-size: 0.85rem;
+    }
+    
+    .metrics-section {
+        padding: 1rem 0.5rem;
+    }
+}
+
 .metrics-grid .metric-card {
     align-self: start;
 }
@@ -474,7 +497,7 @@ table tr:hover td {
 .dataframe table {
     width: 100% !important;
     border-collapse: collapse !important;
-    font-size: 0.95rem !important;
+    font-size: 0.75rem !important;
     table-layout: auto !important;
     background: #ffffff !important;
 }
@@ -621,6 +644,7 @@ table tr:hover td {
     font-family: ui-monospace, SFMono-Regular, Menlo, Monaco, Consolas, "Liberation Mono", "Courier New", monospace !important;
     text-align: left !important;
     white-space: nowrap !important;
+    min-width: 80px !important;
 }
 
 .dataframe td:nth-child(2),