Update app.py

app.py

@@ -8,7 +8,8 @@ from huggingface_hub import login
 import requests 
 import json 
 import base64 # <-- ADDED IMPORT for image handling
-
+import re
+import time
 # Attempt to login using environment token
 try:
     HF_TOKEN = os.environ.get("HUGGINGFACE_TOKEN")
@@ -197,6 +198,30 @@ def get_gaia_file_data_for_task(task_id_for_file_fetch, associated_file_metadata
         print(f"Unexpected error fetching file for task {task_id_for_file_fetch}: {e_gen}")
         return None, None, None
 
+
+def clean_final_answer(raw_text):
+    # Step 1: Extract FINAL ANSWER section if present
+    if "FINAL ANSWER:" in raw_text.upper():
+        match = re.search(r"FINAL ANSWER:\s*(.*)", raw_text, re.IGNORECASE | re.DOTALL)
+        if match:
+            answer = match.group(1).strip()
+        else:
+            answer = raw_text.strip()
+    else:
+        answer = raw_text.strip()
+
+    # Step 2: Remove wrapping quotes if any
+    if (answer.startswith('"') and answer.endswith('"')) or (answer.startswith("'") and answer.endswith("'")):
+        answer = answer[1:-1].strip()
+
+    # Step 3: Remove unwanted units unless explicitly required
+    # You may adjust this depending on the task
+    answer = answer.replace('%', '').replace('$', '').strip()
+
+    # Step 4: Normalize spaces (e.g., for comma-separated answers)
+    answer = re.sub(r'\s*,\s*', ',', answer)
+
+    return answer
 def my_agent_logic(task_id: str, question: str, files_metadata: list = None): # files_metadata is the list from task.get("files")
     """
     Uses the Gemini API, with GAIA-specific prompting and basic file handling, 
@@ -291,17 +316,10 @@ def my_agent_logic(task_id: str, question: str, files_metadata: list = None): #
             result["candidates"][0].get("content") and
             result["candidates"][0]["content"].get("parts") and
             result["candidates"][0]["content"]["parts"][0].get("text")):
-            raw_answer = result["candidates"][0]["content"]["parts"][0]["text"].strip()
             
-            # Remove the "FINAL ANSWER:" prefix if the LLM included it, despite instructions
-            if raw_answer.upper().startswith("FINAL ANSWER:"):
-                agent_computed_answer = raw_answer[len("FINAL ANSWER:"):].strip()
-            else:
-                agent_computed_answer = raw_answer
-            # Further cleaning: sometimes LLMs might still add subtle quotes if the answer is a simple string
-            if len(agent_computed_answer) > 1 and ((agent_computed_answer.startswith('"') and agent_computed_answer.endswith('"')) or \
-               (agent_computed_answer.startswith("'") and agent_computed_answer.endswith("'"))):
-                agent_computed_answer = agent_computed_answer[1:-1]
+            raw_answer = result["candidates"][0]["content"]["parts"][0]["text"].strip()
+            print(f"Raw Gemini output: {raw_answer}")  # Debugging log
+            agent_computed_answer = clean_final_answer(raw_answer)
 
         else:
             print(f"Warning: Unexpected response structure from Gemini API for task {task_id}: {result}")
@@ -364,6 +382,7 @@ def run_agent_on_gaia(profile: gr.OAuthProfile, run_all_questions: bool = True):
             submitted_answer = my_agent_logic(task_id, question, associated_files_metadata)
             log_messages.append(f"Agent's Answer: {submitted_answer}")
             answers_to_submit.append({"task_id": task_id, "submitted_answer": submitted_answer})
+            time.sleep(2) 
         else:
             log_messages.append(f"Skipping malformed task: {task}")
     if not answers_to_submit: