From da19fdbddc49b591a63d1dc1f9136793c519eb0c Mon Sep 17 00:00:00 2001 From: niuyazhe Date: Thu, 2 Sep 2021 21:12:35 +0800 Subject: [PATCH] hotfix(nyz): fix cartpole ppg value buffer sample typo --- dizoo/classic_control/cartpole/entry/cartpole_ppg_main.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/dizoo/classic_control/cartpole/entry/cartpole_ppg_main.py b/dizoo/classic_control/cartpole/entry/cartpole_ppg_main.py index dfdcd169..a44fa076 100644 --- a/dizoo/classic_control/cartpole/entry/cartpole_ppg_main.py +++ b/dizoo/classic_control/cartpole/entry/cartpole_ppg_main.py @@ -66,7 +66,7 @@ def main(cfg, seed=0, max_iterations=int(1e10)): for i in range(cfg.policy.learn.update_per_collect): batch_size = learner.policy.get_attribute('batch_size') policy_data = policy_buffer.sample(batch_size['policy'], learner.train_iter) - value_data = policy_buffer.sample(batch_size['value'], learner.train_iter) + value_data = value_buffer.sample(batch_size['value'], learner.train_iter) if policy_data is not None and value_data is not None: train_data = {'policy': policy_data, 'value': value_data} learner.train(train_data, collector.envstep) -- GitLab