if __name__ == "__main__": parser=argparse.ArgumentParser(description="tournament between agents") #parser.add_argument("num_games", type=int, help="num of paired games playing") parser.add_argument("model_path", help="where the model locates", type=str) parser.add_argument("--wolve_path",default="/home/cgao3/benzene/src/wolve/wolve", help="where is the wolve", type=str) parser.add_argument("--value_net", help="whether it is valuenet model", action="store_true", default=False) parser.add_argument("--verbose", help="verbose?", action="store_true", default=False) args=parser.parse_args() num_games=1000 think_time=1 net_exe=EXE_NN_AGENT_NAME + args.model_path +" 2>/dev/null" EXE_HEX_PATH=args.wolve_path wolve_exe=EXE_HEX_PATH+" 2>/dev/null" wolve=WrapperAgent(wolve_exe, True) net=WrapperAgent(net_exe, True) wolve.sendCommand("param_wolve max_time "+repr(think_time)) white_win_count=0 black_win_count=0 for i in range(num_games): wolve.reconnect() wolve.sendCommand("param_wolve max_time "+ repr(think_time)) wolve.sendCommand("boardsize "+ repr(BOARD_SIZE)) win=run_single_match(net, wolve, False) if win==0: black_win_count += 1 if win==1: white_win_count +=1 print(i+1, "black: ", black_win_count, "white: ", white_win_count) net.sendCommand("close") print("black win ", black_win_count, "white win count ", white_win_count)