hiro/ray
1
0
Fork 0
mirror of https://github.com/vale981/ray synced 2025-03-06 10:31:39 -05:00
Code Issues Projects Releases Packages Wiki Activity Actions

[RLlib]: Trajectory View API: Keep env infos (e.g. for postprocessing callbacks), no matter what. (#13555)

Browse source
This commit is contained in:
Sven Mika 2021-02-09 17:05:26 +01:00 • committed by GitHub
parent 2092b097ea
commit d7301a51f4
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
2 changed files with 4 additions and 4 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

4
rllib/policy/dynamic_tf_policy.py
View file

@ -590,12 +590,12 @@ class DynamicTFPolicy(TFPolicy):
del self._loss_input_dict[key]
# Remove those not needed at all (leave those that are needed
# by Sampler to properly execute sample collection).
# Also always leave DONES and REWARDS, no matter what.
# Also always leave DONES, REWARDS, and INFOS, no matter what.
for key in list(self.view_requirements.keys()):
if key not in all_accessed_keys and key not in [
SampleBatch.EPS_ID, SampleBatch.AGENT_INDEX,
SampleBatch.UNROLL_ID, SampleBatch.DONES,
SampleBatch.REWARDS] and \
SampleBatch.REWARDS, SampleBatch.INFOS] and \
key not in self.model.view_requirements:
# If user deleted this key manually in postprocessing
# fn, warn about it and do not remove from

4
rllib/policy/policy.py
View file

@ -676,12 +676,12 @@ class Policy(metaclass=ABCMeta):
self.view_requirements[key].used_for_training = False
# Remove those not needed at all (leave those that are needed
# by Sampler to properly execute sample collection).
# Also always leave DONES and REWARDS, no matter what.
# Also always leave DONES, REWARDS, INFOS, no matter what.
for key in list(self.view_requirements.keys()):
if key not in all_accessed_keys and key not in [
SampleBatch.EPS_ID, SampleBatch.AGENT_INDEX,
SampleBatch.UNROLL_ID, SampleBatch.DONES,
SampleBatch.REWARDS] and \
SampleBatch.REWARDS, SampleBatch.INFOS] and \
key not in self.model.view_requirements:
# If user deleted this key manually in postprocessing
# fn, warn about it and do not remove from