Update app.py

app.py

@@ -18,7 +18,7 @@ from huggingface_hub import InferenceClient, login
 
 class BasicAgent:
     def __init__(self):
-        login(token=os.environ["HF_TOKEN"])  # Required for gated access
+        login(token=os.environ["HF_TOKEN"])
         
         self.client = InferenceClient(
             model="Qwen/Qwen2-7B-Instruct",
@@ -26,6 +26,11 @@ class BasicAgent:
             timeout=120
         )
         
+        # Verify model access
+        test_response = self.client.text_generation("2+2=", max_new_tokens=10)
+        if "4" not in test_response:
+            raise RuntimeError(f"Model test failed: {test_response}")
+
     def __call__(self, question: str) -> str:
         try:
             prompt = f"""<|im_start|>system
@@ -37,14 +42,15 @@ Answer with ONLY the exact value requested.<|im_end|>
             response = self.client.text_generation(
                 prompt=prompt,
                 temperature=0.01,
-                max_new_tokens=50,
-                stop_sequences=["<|im_end|>"]
+                max_new_tokens=100,
+                stop_sequences=["<|im_end|>"],
+                repetition_penalty=1.1
             )
             return response.split("<|im_start|>assistant")[-1].split("<|im_end|>")[0].strip()
-        except Exception:
+        except Exception as e:
+            print(f"Error processing '{question[:50]}...': {str(e)}")
             return ""
 
-
 def run_and_submit_all( profile: gr.OAuthProfile | None):
     """
     Fetches all questions, runs the BasicAgent on them, submits all answers,