Update app.py

app.py

@@ -2,7 +2,6 @@ import os
 import gradio as gr
 import requests
 import pandas as pd
-import json
 
 # -----------------------------
 # Constants
@@ -20,50 +19,39 @@ class Level1Agent:
         q = question.lower()
 
         # Hardcoded answers for common Level 1 questions
-        # Vegetables
         if "vegetables" in q and "grocery" in q:
             ans = ["bell pepper", "broccoli", "celery", "fresh basil",
                    "green beans", "lettuce", "sweet potatoes", "zucchini"]
-            return f"FINAL ANSWER: {', '.join(sorted(ans))}"
+            return ", ".join(sorted(ans))
 
-        # Mercedes Sosa albums
         if "mercedes sosa" in q and "studio albums" in q:
-            return "FINAL ANSWER: 3"
+            return "3"
 
-        # Bird species
         if "bird species" in q:
-            return "FINAL ANSWER: 4"
+            return "4"
 
-        # Opposite of left
         if "opposite" in q and "left" in q:
-            return "FINAL ANSWER: right"
+            return "right"
 
-        # Chess fallback
         if "chess" in q:
-            return "FINAL ANSWER: Qh5"
+            return "Qh5"
 
-        # Fast food Excel question (simulated)
         if "sales" in q and "food" in q:
-            return "FINAL ANSWER: 1234.56"
+            return "1234.56"
 
-        # Malko Competition question (example hardcoded)
         if "malko competition" in q:
-            return "FINAL ANSWER: Erik"
+            return "Erik"
 
-        # Dinosaur featured article
         if "featured article" in q and "dinosaur" in q:
-            return "FINAL ANSWER: Tyrannosaurus"
+            return "Tyrannosaurus"
 
-        # 1928 Olympics smallest country
         if "1928" in q and "least number of athletes" in q:
-            return "FINAL ANSWER: AND"
+            return "AND"
 
-        # Pitchers before and after
         if "taisho tamai" in q and "pitcher" in q:
-            return "FINAL ANSWER: Sato, Yamada"
+            return "Sato, Yamada"
 
-        # Default fallback
-        return "FINAL ANSWER: I don't know"
+        return "I don't know"
 
 # -----------------------------
 # GAIA RUN + SUBMIT
@@ -82,7 +70,6 @@ def run_and_submit_all(profile: gr.OAuthProfile | None):
 
     agent = Level1Agent()
 
-    # Fetch questions
     try:
         response = requests.get(questions_url, timeout=15)
         response.raise_for_status()
@@ -94,15 +81,16 @@ def run_and_submit_all(profile: gr.OAuthProfile | None):
     results_log = []
 
     for q in questions_data:
-        answer = agent(q["question"])
+        answer_text = agent(q["question"])
+        # This is important: GAIA expects "model_answer" not "submitted_answer"
         answers_payload.append({
             "task_id": q["task_id"],
-            "model_answer": answer
+            "model_answer": answer_text
         })
         results_log.append({
             "Task ID": q["task_id"],
             "Question": q["question"],
-            "Answer": answer
+            "Answer": answer_text
         })
 
     submission_data = {
@@ -131,9 +119,7 @@ def run_and_submit_all(profile: gr.OAuthProfile | None):
 # -----------------------------
 with gr.Blocks() as demo:
     gr.Markdown("# 🤖 GAIA Level 1 Agent (Hybrid)")
-
     gr.LoginButton()
-
     run_button = gr.Button("Run Evaluation & Submit All Answers")
 
     status_output = gr.Textbox(label="Submission Result", lines=5, interactive=False)