def run_agents(agent1, agent2, episodes=100, verbose=False, train=True):
agent1wins = 0
agent2wins = 0
draws = 0
for episode in range(episodes):
game = TicTacToe(**game_settings)
res = STEP_RESULT.NONE
agent1_team = X
agent2_team = O
if episode % 2 == 1:
agent1_team = X
agent2_team = O
for i in range(game_settings["size"]**2):
team_this_turn = i % 2
state = game.state(team_this_turn)
action = None
if (i % 2 == agent1_team):
action = agent1.step(state, train)
else:
action = agent2.step(state, train)
_, _, res, _ = game.apply(action)
if (res == STEP_RESULT.DRAW):
draws += 1
break
elif res == STEP_RESULT.X_WIN:
if agent1_team == X:
agent1wins += 1
else:
agent2wins += 1
break
elif res == STEP_RESULT.O_WIN:
if agent1_team == O:
agent1wins += 1
else:
agent2wins += 1
break
state = game.state(agent1_team)
agent1.step(state, train)
state = game.state(agent2_team)
agent2.step(state, train)
if verbose:
print(
"episode {} - Score: Agent 1 {} - Agent 2 {} - Draws {} - Agent2 Pts - {}"
.format(episode + 1, agent1wins, agent2wins, draws,
agent2wins - agent1wins - draws))
