wassname/ray
1
0
Fork 0
mirror of https://github.com/wassname/ray.git synced 2026-09-10 12:38:43 +08:00
Code Issues Packages Projects Releases Wiki Activity

[RLlib] Add 2 Transformer learning test cases on StatelessCartPole (PPO and IMPALA). (#8624)

Browse Source
This commit is contained in:
Sven Mika
2020-05-27 10:19:47 +02:00
committed by GitHub
parent b0bb0584fb
commit 0422e9c5a8
14 changed files with 133 additions and 33 deletions
Download Patch File Download Diff File Expand all files Collapse all files
rllib/examples/models/custom_loss_model.py
+3 -2
View File
@@ -38,7 +38,8 @@ class CustomLossModel(TFModelV2):
@override(ModelV2)
def custom_loss(self, policy_loss, loss_inputs):
# Create a new input reader per worker.
reader = JsonReader(self.model_config["custom_options"]["input_files"])
reader = JsonReader(
self.model_config["custom_model_config"]["input_files"])
input_ops = reader.tf_input_ops()
# Define a secondary loss by building a graph copy with weight sharing.
@@ -80,7 +81,7 @@ class DeprecatedCustomLossModelV1(Model):
def custom_loss(self, policy_loss, loss_inputs):
# create a new input reader per worker
reader = JsonReader(self.options["custom_options"]["input_files"])
reader = JsonReader(self.options["custom_model_config"]["input_files"])
input_ops = reader.tf_input_ops()
# define a secondary loss by building a graph copy with weight sharing