Unity-Technologies · ervteng · Apr 29, 2020 · Apr 14, 2020 · Apr 14, 2020 · Apr 15, 2020
diff --git a/Project/Recordings/movie.mp4 b/Project/Recordings/movie.mp4
diff --git a/com.unity.ml-agents/CHANGELOG.md b/com.unity.ml-agents/CHANGELOG.md
@@ -36,7 +36,10 @@ and this project adheres to
  communication between Unity and the Python process.
 - The obsolete `Agent` methods `GiveModel`, `Done`, `InitializeAgent`,
  `AgentAction` and `AgentReset` have been removed.
-- The GhostTrainer has been extended to support asymmetric games and the asymmetric example environment Strikers Vs. Goalie has been added.
+- The GhostTrainer has been extended to support asymmetric games and the asymmetric example environment Strikers Vs. Goalie has been added.
+- Curriculum and Parameter Randomization configurations have been merged
+ into the main training configuration file. Note that this means training
+ configuration files are environment-specific. (#3791)
 
 ### Minor Changes
 

diff --git a/config/3dball_randomize.yaml b/config/3dball_randomize.yaml
diff --git a/config/curricula/soccer.yaml b/config/curricula/soccer.yaml
diff --git a/config/curricula/test.yaml b/config/curricula/test.yaml
diff --git a/config/curricula/wall_jump.yaml b/config/curricula/wall_jump.yaml
diff --git a/config/gail_config.yaml b/config/gail_config.yaml
diff --git a/config/imitation/CrawlerStatic.yaml b/config/imitation/CrawlerStatic.yaml
@@ -0,0 +1,29 @@
+behaviors:
+ CrawlerStatic:
+ trainer: ppo
+ batch_size: 2024
+ beta: 0.005
+ buffer_size: 20240
+ epsilon: 0.2
+ hidden_units: 512
+ lambd: 0.95
+ learning_rate: 0.0003
+ max_steps: 1e7
+ memory_size: 256
+ normalize: true
+ num_epoch: 3
+ num_layers: 3
+ time_horizon: 1000
+ sequence_length: 64
+ summary_freq: 30000
+ use_recurrent: false
+ reward_signals:
+ gail:
+ strength: 1.0
+ gamma: 0.99
+ encoding_size: 128
+ demo_path: Project/Assets/ML-Agents/Examples/Crawler/Demos/ExpertCrawlerSta.demo
+ behavioral_cloning:
+ demo_path: Project/Assets/ML-Agents/Examples/Crawler/Demos/ExpertCrawlerSta.demo
+ strength: 0.5
+ steps: 50000
diff --git a/config/imitation/FoodCollector.yaml b/config/imitation/FoodCollector.yaml
@@ -0,0 +1,29 @@
+behaviors:
+ FoodCollector:
+ trainer: ppo
+ batch_size: 64
+ beta: 0.005
+ buffer_size: 10240
+ epsilon: 0.2
+ hidden_units: 128
+ lambd: 0.95
+ learning_rate: 0.0003
+ max_steps: 2.0e6
+ memory_size: 256
+ normalize: false
+ num_epoch: 3
+ num_layers: 2
+ time_horizon: 64
+ sequence_length: 32
+ summary_freq: 10000
+ use_recurrent: false
+ reward_signals:
+ gail:
+ strength: 0.1
+ gamma: 0.99
+ encoding_size: 128
+ demo_path: Project/Assets/ML-Agents/Examples/FoodCollector/Demos/ExpertFood.demo
+ behavioral_cloning:
+ demo_path: Project/Assets/ML-Agents/Examples/FoodCollector/Demos/ExpertFood.demo
+ strength: 1.0
+ steps: 0
diff --git a/config/imitation/Hallway.yaml b/config/imitation/Hallway.yaml
@@ -0,0 +1,28 @@
+behaviors:
+ Hallway:
+ trainer: ppo
+ batch_size: 128
+ beta: 0.01
+ buffer_size: 1024
+ epsilon: 0.2
+ hidden_units: 128
+ lambd: 0.95
+ learning_rate: 0.0003
+ max_steps: 1.0e7
+ memory_size: 256
+ normalize: false
+ num_epoch: 3
+ num_layers: 2
+ time_horizon: 64
+ sequence_length: 64
+ summary_freq: 10000
+ use_recurrent: true
+ reward_signals:
+ extrinsic:
+ strength: 1.0
+ gamma: 0.99
+ gail:
+ strength: 0.1
+ gamma: 0.99
+ encoding_size: 128
+ demo_path: Project/Assets/ML-Agents/Examples/Hallway/Demos/ExpertHallway.demo
diff --git a/config/imitation/PushBlock.yaml b/config/imitation/PushBlock.yaml
@@ -0,0 +1,25 @@
+behaviors:
+ PushBlock:
+ trainer: ppo
+ batch_size: 128
+ beta: 0.01
+ buffer_size: 2048
+ epsilon: 0.2
+ hidden_units: 256
+ lambd: 0.95
+ learning_rate: 0.0003
+ max_steps: 1.5e7
+ memory_size: 256
+ normalize: false
+ num_epoch: 3
+ num_layers: 2
+ time_horizon: 64
+ sequence_length: 64
+ summary_freq: 60000
+ use_recurrent: false
+ reward_signals:
+ gail:
+ strength: 1.0
+ gamma: 0.99
+ encoding_size: 128
+ demo_path: Project/Assets/ML-Agents/Examples/PushBlock/Demos/ExpertPush.demo
diff --git a/config/imitation/Pyramids.yaml b/config/imitation/Pyramids.yaml
@@ -0,0 +1,36 @@
+behaviors:
+ Pyramids:
+ trainer: ppo
+ batch_size: 128
+ beta: 0.01
+ buffer_size: 2048
+ epsilon: 0.2
+ hidden_units: 512
+ lambd: 0.95
+ learning_rate: 0.0003
+ max_steps: 1.0e7
+ memory_size: 256
+ normalize: false
+ num_epoch: 3
+ num_layers: 2
+ time_horizon: 128
+ sequence_length: 64
+ summary_freq: 30000
+ use_recurrent: false
+ reward_signals:
+ extrinsic:
+ strength: 1.0
+ gamma: 0.99
+ curiosity:
+ strength: 0.02
+ gamma: 0.99
+ encoding_size: 256
+ gail:
+ strength: 0.01
+ gamma: 0.99
+ encoding_size: 128
+ demo_path: Project/Assets/ML-Agents/Examples/Pyramids/Demos/ExpertPyramid.demo
+ behavioral_cloning:
+ demo_path: Project/Assets/ML-Agents/Examples/Pyramids/Demos/ExpertPyramid.demo
+ strength: 0.5
+ steps: 150000
diff --git a/config/ppo/3DBall.yaml b/config/ppo/3DBall.yaml
@@ -0,0 +1,25 @@
+behaviors:
+ 3DBall:
+ trainer: ppo
+ batch_size: 64
+ beta: 0.001
+ buffer_size: 12000
+ epsilon: 0.2
+ hidden_units: 128
+ lambd: 0.99
+ learning_rate: 0.0003
+ learning_rate_schedule: linear
+ max_steps: 5.0e5
+ memory_size: 128
+ normalize: true
+ num_epoch: 3
+ num_layers: 2
+ time_horizon: 1000
+ sequence_length: 64
+ summary_freq: 12000
+ use_recurrent: false
+ vis_encode_type: simple
+ reward_signals:
+ extrinsic:
+ strength: 1.0
+ gamma: 0.99
diff --git a/config/ppo/3DBallHard.yaml b/config/ppo/3DBallHard.yaml
@@ -0,0 +1,25 @@
+behaviors:
+ 3DBallHard:
+ trainer: ppo
+ batch_size: 1200
+ beta: 0.001
+ buffer_size: 12000
+ epsilon: 0.2
+ hidden_units: 128
+ lambd: 0.95
+ learning_rate: 0.0003
+ learning_rate_schedule: linear
+ max_steps: 5.0e6
+ memory_size: 128
+ normalize: true
+ num_epoch: 3
+ num_layers: 2
+ time_horizon: 1000
+ sequence_length: 64
+ summary_freq: 12000
+ use_recurrent: false
+ vis_encode_type: simple
+ reward_signals:
+ extrinsic:
+ strength: 1.0
+ gamma: 0.995