5e76b6981d0b48def00d4fa97caec24accf402fd,contents/5.2_Prioritized_Replay_DQN/RL_brain.py,Memory,sample,#Memory#,116

Before Change

        self.beta = np.min([1, self.beta + self.beta_increment_per_sampling])  // max = 1

        min_prob = np.min(self.tree.tree[-self.tree.capacity:]) / self.tree.root_priority
        maxiwi = np.power(self.tree.capacity * min_prob, -self.beta)  // for later normalizing ISWeights
        for i in range(n):
            a = segment * i
            b = segment * (i + 1)
            lower_bound = np.random.uniform(a, b)
            idx, p, data = self.tree.get_leaf(lower_bound)
            prob = p / self.tree.root_priority
            ISWeights.append(self.tree.capacity * prob)
            batch_idx.append(idx)
            batch_memory.append(data)

        ISWeights = np.vstack(ISWeights)
        ISWeights = np.power(ISWeights, -self.beta) / maxiwi  // normalize
        return batch_idx, np.vstack(batch_memory), ISWeights

    def update(self, idx, error):

After Change

            v = np.random.uniform(a, b)
            idx, p, data = self.tree.get_leaf(v)
            prob = p / self.tree.total_p
            ISWeights[i, 0] = np.power(prob/min_prob, -self.beta)
            b_idx[i], b_memory[i, :] = idx, data
        return b_idx, b_memory, ISWeights

Instances