Update app.py

app.py

@@ -88,14 +88,18 @@ def run_evaluation(num_samples, progress=gr.Progress()):
     
     progress(0.1, desc="Loading CUAD dataset...")
     
-    # Load dataset
+    # Load dataset - use QA format version (JSON, no PDFs)
     try:
-        dataset = load_dataset("cuad", trust_remote_code=True, token=hf_token)
+        # Try the QA-specific version first (much faster, JSON format)
+        dataset = load_dataset("theatticusproject/cuad-qa", trust_remote_code=True, token=hf_token)
         test_data = dataset["test"]
+        print(f"✓ Loaded CUAD-QA dataset with {len(test_data)} samples")
     except Exception as e:
         try:
-            dataset = load_dataset("theatticusproject/cuad", trust_remote_code=True, token=hf_token)
-            test_data = dataset["test"]
+            # Fallback to original but limit to avoid PDF downloads
+            dataset = load_dataset("cuad", split="test[:1000]", trust_remote_code=True, token=hf_token)
+            test_data = dataset
+            print(f"✓ Loaded CUAD dataset with {len(test_data)} samples")
         except Exception as e2:
             return f"❌ Error loading dataset: {e2}", pd.DataFrame(), None