wassname/ray
1
0
Fork 0
mirror of https://github.com/wassname/ray.git synced 2026-07-31 12:41:01 +08:00
Code Issues Packages Projects Releases Wiki Activity

[rllib] more user-friendly Optimizer signature + compute_apply (#2335)

Browse Source 
* Move signature of optimizers

* fix

* expose compute_apply for policy_graphs

* dictionaries and such

* test for multiagent
This commit is contained in:
Richard Liaw
2018-07-07 12:08:49 -07:00
committed by GitHub
parent e3534c46df
commit e32aed8717
10 changed files with 35 additions and 21 deletions
Download Patch File Download Diff File Expand all files Collapse all files
python/ray/rllib/agents/ppo/ppo.py
+2 -2
View File
@@ -74,11 +74,11 @@ class PPOAgent(Agent):
{"num_cpus": self.config["num_cpus_per_worker"],
"num_gpus": self.config["num_gpus_per_worker"]})
self.optimizer = LocalMultiGPUOptimizer(
self.local_evaluator, self.remote_evaluators,
{"sgd_batch_size": self.config["sgd_batchsize"],
"sgd_stepsize": self.config["sgd_stepsize"],
"num_sgd_iter": self.config["num_sgd_iter"],
"timesteps_per_batch": self.config["timesteps_per_batch"]},
self.local_evaluator, self.remote_evaluators)
"timesteps_per_batch": self.config["timesteps_per_batch"]})
def _train(self):
def postprocess_samples(batch):