fix bug

app.py

@@ -422,11 +422,10 @@ with app:
             gr.Markdown("**Note:** Progress percentage will be shown during evaluation.")
             gr.Markdown("**Tip:** Limit the number of samples for faster evaluation during testing.")
             
-            max_samples_input = gr.Number(
+            max_samples_input = gr.Textbox(
                 label="Maximum samples to evaluate (leave empty for full dataset)",
-                value=None,
-                minimum=1,
-                precision=0,
+                value="",
+                placeholder="Enter a number or leave empty for full dataset",
                 info="Set a limit to evaluate faster. Leave empty to evaluate on the full dataset."
             )
             
@@ -437,9 +436,18 @@ with app:
             )
             eval_output = gr.Markdown(label="Evaluation Results")
             
-            def run_evaluation(max_samples):
+            def run_evaluation(max_samples_str):
                 """Run evaluation and return result."""
-                max_samples_int = int(max_samples) if max_samples is not None and max_samples > 0 else None
+                # Handle empty string, None, or invalid values as "use full dataset"
+                max_samples_int = None
+                if max_samples_str and max_samples_str.strip():
+                    try:
+                        max_samples_int = int(max_samples_str.strip())
+                        if max_samples_int < 1:
+                            max_samples_int = None
+                    except (ValueError, TypeError):
+                        max_samples_int = None
+                
                 result = evaluate_test_set(max_samples=max_samples_int)
                 return result