wassname/DeepRL
1
0
Fork 0
mirror of https://github.com/wassname/DeepRL.git synced 2026-09-09 11:13:47 +08:00
Code Issues Packages Projects Releases Wiki Activity

tensorboard logging for ppo reward

Browse Source
This commit is contained in:
Mike Clark
2018-02-02 12:09:57 +08:00
committed by GitHub
parent 823bda6fee
commit 8274147aff
1 changed files with 16 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
async_worker/ppo.py
+16
View File
@@ -78,6 +78,22 @@ class ProximalPolicyOptimization:
batched_steps += 1
episode_length += 1
# tensorboard logging
total_steps = config.total_steps.value + batched_steps
suffix = 'test_' if deterministic else ''
if action.squeeze().ndim == 0:
config.logger.scalar_summary(suffix + 'action', action, total_steps)
config.logger.scalar_summary(suffix + 'mean', mean, total_steps)
config.logger.scalar_summary(suffix + 'std', std, total_steps)
else:
config.logger.histo_summary(suffix + 'action', action, total_steps)
config.logger.histo_summary(suffix + 'mean', mean, total_steps)
config.logger.histo_summary(suffix + 'std', std, total_steps)
config.logger.scalar_summary(suffix + 'value', value, total_steps)
config.logger.scalar_summary(suffix + 'reward', reward, total_steps)
for key in info:
config.logger.scalar_summary('info_' + key, info[key], total_steps)
reward = self.reward_normalizer(reward)
rewards.append(reward)