d1fe177a4f3057e9e8b71ab820dbe5fe2a538276,ch07/04_dqn_noisy_net.py,,unpack_batch,#Any#,37

Before Change

def unpack_batch(batch):
    states = [exp[0].state for exp in batch]
    next_states = [exp[-1].state for exp in batch]
    actions = [exp[0].action for exp in batch]
    rewards = [exp[0].reward for exp in batch]
    dones = [exp[0].done for exp in batch]
    return np.array(states, copy=False), np.array(actions), np.array(rewards, dtype=np.float32), \
           np.array(dones, dtype=np.uint8), np.array(next_states, copy=False)

After Change

def unpack_batch(batch):
    states, actions, rewards, dones, last_states = [], [], [], [], []
    for exp in batch:
        state = np.array(exp.state, copy=False)
        states.append(state)
        actions.append(exp.action)
        rewards.append(exp.reward)
        dones.append(exp.last_state is None)
        if exp.last_state is None:
            last_states.append(state)       // the result will be masked anyway
        else:
            last_states.append(np.array(exp.last_state, copy=False))
    return np.array(states, copy=False), np.array(actions), np.array(rewards, dtype=np.float32), \
           np.array(dones, dtype=np.uint8), np.array(last_states, copy=False)

Instances