Upload stratego\benchmarking\plot_metrics.py with huggingface_hub

stratego//benchmarking//plot_metrics.py

@@ -0,0 +1,90 @@
+# stratego/benchmarking/plot_metrics.py
+
+import pandas as pd
+import matplotlib.pyplot as plt
+import sys
+
+
+def plot_from_csv(csv_path: str, rolling_window: int = 3):
+    df = pd.read_csv(csv_path)
+
+    games = range(len(df))
+
+    # ===============================
+    # 1. GAME LENGTH PER GAME
+    # ===============================
+    plt.figure()
+    plt.plot(games, df["turns"], marker="o")
+    plt.title("Game Length per Game")
+    plt.xlabel("Game index")
+    plt.ylabel("Turns")
+    plt.grid(True)
+    plt.show()
+
+    # ===============================
+    # 2. INVALID MOVES PER GAME
+    # ===============================
+    plt.figure()
+    plt.plot(games, df["invalid_moves_p0"], label="P0 Invalid Moves", marker="o")
+    plt.plot(games, df["invalid_moves_p1"], label="P1 Invalid Moves", marker="o")
+    plt.title("Invalid Moves per Game")
+    plt.xlabel("Game index")
+    plt.ylabel("Invalid moves")
+    plt.legend()
+    plt.grid(True)
+    plt.show()
+
+    # ===============================
+    # 3. ROLLING AVERAGE (STALLING)
+    # ===============================
+    df["rolling_turns"] = df["turns"].rolling(window=rolling_window).mean()
+
+    plt.figure()
+    plt.plot(games, df["turns"], alpha=0.3, label="Raw turns")
+    plt.plot(games, df["rolling_turns"], linewidth=3, label=f"Rolling avg ({rolling_window})")
+    plt.title("Game Stalling (Rolling Average of Turns)")
+    plt.xlabel("Game index")
+    plt.ylabel("Turns")
+    plt.legend()
+    plt.grid(True)
+    plt.show()
+
+    # ===============================
+    # 4. TERMINATION REASONS
+    # ===============================
+    termination_counts = df["game_end_reason"].value_counts()
+
+    plt.figure()
+    termination_counts.plot(kind="bar")
+    plt.title("Game Termination Reasons")
+    plt.xlabel("Reason")
+    plt.ylabel("Number of games")
+    plt.grid(axis="y")
+    plt.show()
+
+    # ===============================
+    # 5. CUMULATIVE WIN RATE
+    # ===============================
+    p0_wins = (df["winner"] == 0).cumsum()
+    p1_wins = (df["winner"] == 1).cumsum()
+
+    win_rate_p0 = p0_wins / (games + 1)
+    win_rate_p1 = p1_wins / (games + 1)
+
+    plt.figure()
+    plt.plot(games, win_rate_p0, label="P0 Win Rate")
+    plt.plot(games, win_rate_p1, label="P1 Win Rate")
+    plt.title("Cumulative Win Rate")
+    plt.xlabel("Game index")
+    plt.ylabel("Win rate")
+    plt.legend()
+    plt.grid(True)
+    plt.show()
+
+
+if __name__ == "__main__":
+    if len(sys.argv) < 2:
+        print("Usage: python plot_metrics.py <benchmark_csv>")
+        sys.exit(1)
+
+    plot_from_csv(sys.argv[1])