MCPcopy Create free account
hub / github.com/AgentMaker/Paddle-RLBooks / sample

Method sample

ddpg/replay_memory.py:14–31  ·  view source on GitHub ↗
(self, batch_size)

Source from the content-addressed store, hash-verified

12 self.buffer.append(exp)
13
14 def sample(self, batch_size):
15 mini_batch = random.sample(self.buffer, batch_size)
16 batch_state, batch_action, batch_reward, batch_next_state, batch_done = [], [], [], [], []
17
18 for experience in mini_batch:
19 s, a, r, s_p, done = experience
20 batch_state.append(s)
21 batch_action.append(a)
22 batch_reward.append(r)
23 batch_next_state.append(s_p)
24 batch_done.append(done)
25 batch_state = paddle.to_tensor(batch_state, dtype='float32')
26 batch_action = paddle.to_tensor(batch_action, dtype='float32')
27 batch_reward = paddle.to_tensor(batch_reward, dtype='float32')
28 batch_next_state = paddle.to_tensor(batch_next_state, dtype='float32')
29 batch_done = paddle.to_tensor(batch_done, dtype='float32')
30
31 return batch_state, batch_action, batch_reward, batch_next_state, batch_done
32
33 def __len__(self):
34 return len(self.buffer)

Callers 1

trainFunction · 0.45

Calls 1

appendMethod · 0.45

Tested by

no test coverage detected