wassname/DeepRL
1
0
Fork 0
mirror of https://github.com/wassname/DeepRL.git synced 2026-08-30 11:14:19 +08:00
Code Issues Packages Projects Releases Wiki Activity

log mean of noise

Browse Source
This commit is contained in:
Mike C
2018-02-26 11:42:36 +08:00
parent 96b9860125
commit c8a36198d4
1 changed files with 9 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files
agent/DDPG_agent.py
+9 -10
View File
@@ -79,19 +79,18 @@ class DDPGAgent:
next_state = self.state_normalizer(next_state)
total_reward += reward
assert np.isfinite(action), 'action should be finite'
# tensorboard logging
suffix = 'test_' if deterministic else ''
if action.squeeze().ndim == 0:
config.logger.scalar_summary(suffix + 'action', action, self.total_steps)
config.logger.scalar_summary(suffix + 'noise', noise, self.total_steps)
else:
config.logger.histo_summary(suffix + 'action', action, self.total_steps)
config.logger.histo_summary(suffix + 'noise', noise, self.total_steps)
config.logger.scalar_summary(suffix + 'reward', reward, self.total_steps)
for key in info:
config.logger.scalar_summary('info_' + key, info[key], self.total_steps)
if deterministic or ((steps % 10) == 0):
# it will log to much data if we log every step
if action.squeeze().ndim == 0:
config.logger.scalar_summary(suffix + 'action', action, self.total_steps)
else:
config.logger.histo_summary(suffix + 'action', action, self.total_steps)
config.logger.scalar_summary(suffix + 'noise', np.mean(noise), self.total_steps)
for key in info:
config.logger.scalar_summary('info_' + key, info[key], self.total_steps)
reward = self.reward_normalizer(reward) * config.reward_scaling