Update app.py

app.py

@@ -61,17 +61,25 @@ def extract_leaderboard_stats(game_name: str) -> pd.DataFrame:
         conn = sqlite3.connect(db_file)
         agent_type, model_name = extract_agent_info(db_file)
         
+        # Skip random agent rows
+        if agent_type == "random":
+            conn.close()
+            continue
+        
         if game_name == "Total Performance":
             query = "SELECT game_name, COUNT(DISTINCT episode) AS games_played, " \
                     "SUM(reward) AS total_rewards " \
-                    "FROM game_results GROUP BY game_name"
-            df = pd.read_sql_query(query, conn)
+                    "FROM game_results WHERE game_name = ? GROUP BY game_name"
+            df = pd.read_sql_query(query, conn, params=(game_name,))
         else:
             query = "SELECT COUNT(DISTINCT episode) AS games_played, " \
                     "SUM(reward) AS total_rewards " \
                     "FROM game_results WHERE game_name = ?"
             df = pd.read_sql_query(query, conn, params=(game_name,))
         
+        # Ensure rewards are correctly summed per agent, not duplicated
+        df["total_rewards"] = df["total_rewards"].fillna(0).astype(float) / 2
+        
         # Fetch average generation time from moves table
         gen_time_query = """
             SELECT AVG(generation_time) FROM moves WHERE game_name = ?
@@ -80,20 +88,22 @@ def extract_leaderboard_stats(game_name: str) -> pd.DataFrame:
         
         # Calculate win rate against random bot using moves table
         vs_random_query = """
-            SELECT COUNT(*) FROM moves 
-            WHERE game_name = ? AND opponent = 'random_None' AND action IS NOT NULL
+            SELECT COUNT(DISTINCT episode) FROM game_results gr
+            JOIN moves m ON gr.game_name = m.game_name AND gr.episode = m.episode
+            WHERE gr.game_name = ? AND m.opponent = 'random_None' AND gr.reward > 0
         """
         total_vs_random_query = """
-            SELECT COUNT(*) FROM moves 
-            WHERE game_name = ? AND opponent = 'random_None'
+            SELECT COUNT(DISTINCT episode) FROM game_results gr
+            JOIN moves m ON gr.game_name = m.game_name AND gr.episode = m.episode
+            WHERE gr.game_name = ? AND m.opponent = 'random_None'
         """
         wins_vs_random = conn.execute(vs_random_query, (game_name,)).fetchone()[0] or 0
         total_vs_random = conn.execute(total_vs_random_query, (game_name,)).fetchone()[0] or 0
         vs_random_rate = (wins_vs_random / total_vs_random * 100) if total_vs_random > 0 else 0
         
-        df["agent_name"] = model_name
+        df.insert(0, "agent_name", model_name)  # Ensure agent_name is the first column
         df["agent_type"] = agent_type
-        df["avg_generation_time"] = round(avg_gen_time, 2)
+        df["avg_generation_time (sec)"] = round(avg_gen_time, 3)
         df["vs_random"] = round(vs_random_rate, 2)
         
         all_stats.append(df)
@@ -102,7 +112,7 @@ def extract_leaderboard_stats(game_name: str) -> pd.DataFrame:
     leaderboard_df = pd.concat(all_stats, ignore_index=True) if all_stats else pd.DataFrame()
     
     if leaderboard_df.empty:
-        leaderboard_df = pd.DataFrame(columns=["LLM Model", "# games", "total rewards", "avg gen time", "win-rate", "vs Random"])
+        leaderboard_df = pd.DataFrame(columns=["agent_name", "# games", "total rewards", "avg_generation_time (sec)", "win-rate", "vs_random"])
     
     return leaderboard_df
 
@@ -120,7 +130,7 @@ with gr.Blocks() as interface:
         gr.Markdown("# LLM Model Leaderboard\nTrack performance across different games!")
         available_games = get_available_games()
         leaderboard_game_dropdown = gr.Dropdown(available_games, label="Select Game", value="Total Performance")
-        leaderboard_table = gr.Dataframe(headers=["LLM Model", "# games", "total rewards", "avg gen time", "win-rate", "vs Random"])
+        leaderboard_table = gr.Dataframe(headers=["agent_name", "# games", "total rewards", "avg_generation_time (sec)", "win-rate", "vs_random"])
         generate_button = gr.Button("Generate Leaderboard JSON")
         download_component = gr.File(label="Download Leaderboard JSON")
         refresh_button = gr.Button("Refresh Leaderboard")
@@ -129,4 +139,4 @@ with gr.Blocks() as interface:
         refresh_button.click(extract_leaderboard_stats, inputs=[leaderboard_game_dropdown], outputs=[leaderboard_table])
         generate_button.click(generate_leaderboard_json, outputs=[download_component])
 
-interface.launch()
+interface.launch()