batch_size=64,
layer1_size=256,
layer2_size=128,
n_actions=3)
#agent.load_models()
# np.random.seed(1)
score_history = []
for i in range(50):
obs = env.reset()
done = False
score = 0
while not done:
act = agent.choose_action(obs)
print(act)
new_state, reward, done, info = env.step(act)
agent.remember(obs, act, reward, new_state, int(done))
agent.learn()
score += reward
obs = new_state
#env.render()
score_history.append(score)
if i % 10 == 0:
agent.save_models()
env.render()
print('episode ', i, 'score %.2f' % score,
'trailing 25 games avg %.3f' % np.mean(score_history[-25:]))
next_state, reward, done, _ = env.step(action)
agent.memory.add(state, action, reward, next_state, done)
state = next_state
score += reward
agent.optimize()
score_history.append(score)
avg_score: float = np.mean(score_history[-100:])
avg_history.append(avg_score)
if avg_score > best_score:
best_score = avg_score
agent.save_models(data_path)
print(f'Episode:{i}'
f'\t ACC. Rewards: {score:3.2f}'
f'\t AVG. Rewards: {avg_score:3.2f}'
f'\t *** MODEL SAVED! ***')
else:
print(f'Episode:{i}'
f'\t ACC. Rewards: {score:3.2f}'
f'\t AVG. Rewards: {avg_score:3.2f}')
episode_info = {
'Episode': i,
'Total Episodes': n_games,
'Epidosic Summed Rewards': score,
'Moving Mean of Episodic Rewards': avg_score
}