Merge remote-tracking branch 'origin/master' into release_1_to_master

config/sac/WormDynamic.yaml

+behaviors:
+  WormDynamic:
+    trainer: sac
+    batch_size: 256
+    buffer_size: 500000
+    buffer_init_steps: 0
+    hidden_units: 512
+    init_entcoef: 1.0
+    learning_rate: 0.0003
+    learning_rate_schedule: constant
+    max_steps: 5e6
+    memory_size: 128
+    normalize: true
+    steps_per_update: 20
+    num_layers: 3
+    time_horizon: 1000
+    sequence_length: 64
+    summary_freq: 30000
+    tau: 0.005
+    use_recurrent: false
+    vis_encode_type: simple
+    reward_signals:
+      extrinsic:
+        strength: 1.0
+        gamma: 0.995

config/sac/WormStatic.yaml

+behaviors:
+  WormStatic:
+    trainer: sac
+    batch_size: 256
+    buffer_size: 500000
+    buffer_init_steps: 2000
+    hidden_units: 512
+    init_entcoef: 1.0
+    learning_rate: 0.0003
+    learning_rate_schedule: constant
+    max_steps: 3e6
+    memory_size: 128
+    normalize: true
+    steps_per_update: 20
+    num_layers: 3
+    time_horizon: 1000
+    sequence_length: 64
+    summary_freq: 30000
+    tau: 0.005
+    use_recurrent: false
+    vis_encode_type: simple
+    reward_signals:
+      extrinsic:
+        strength: 1.0
+        gamma: 0.995

config/sac_trainer_config.yaml

-default:
-    trainer: sac
-    batch_size: 128
-    buffer_size: 50000
-    buffer_init_steps: 0
-    hidden_units: 128
-    init_entcoef: 1.0
-    learning_rate: 3.0e-4
-    learning_rate_schedule: constant
-    max_steps: 5.0e5
-    memory_size: 128
-    normalize: false
-    steps_per_update: 10
-    num_layers: 2
-    time_horizon: 64
-    sequence_length: 64
-    summary_freq: 10000
-    tau: 0.005
-    use_recurrent: false
-    vis_encode_type: simple
-    reward_signals:
-        extrinsic:
-            strength: 1.0
-            gamma: 0.99
-
-FoodCollector:
-    normalize: false
-    batch_size: 256
-    buffer_size: 500000
-    max_steps: 2.0e6
-    init_entcoef: 0.05
-
-Bouncer:
-    normalize: true
-    max_steps: 1.0e6
-    num_layers: 2
-    hidden_units: 64
-    summary_freq: 20000
-
-PushBlock:
-    max_steps: 2e6
-    init_entcoef: 0.05
-    hidden_units: 256
-    summary_freq: 100000
-    time_horizon: 64
-    num_layers: 2
-
-SmallWallJump:
-    max_steps: 5e6
-    hidden_units: 256
-    summary_freq: 20000
-    time_horizon: 128
-    init_entcoef: 0.1
-    num_layers: 2
-    normalize: false
-
-BigWallJump:
-    max_steps: 2e7
-    hidden_units: 256
-    summary_freq: 20000
-    time_horizon: 128
-    num_layers: 2
-    init_entcoef: 0.1
-    normalize: false
-
-Striker:
-    max_steps: 5.0e6
-    learning_rate: 1e-3
-    hidden_units: 256
-    summary_freq: 20000
-    time_horizon: 128
-    init_entcoef: 0.1
-    num_layers: 2
-    normalize: false
-
-Goalie:
-    max_steps: 5.0e6
-    learning_rate: 1e-3
-    hidden_units: 256
-    summary_freq: 20000
-    time_horizon: 128
-    init_entcoef: 0.1
-    num_layers: 2
-    normalize: false
-
-Pyramids:
-    summary_freq: 30000
-    time_horizon: 128
-    batch_size: 128
-    buffer_init_steps: 10000
-    buffer_size: 500000
-    hidden_units: 256
-    num_layers: 2
-    init_entcoef: 0.01
-    max_steps: 1.0e7
-    sequence_length: 16
-    tau: 0.01
-    use_recurrent: false
-    reward_signals:
-        extrinsic:
-            strength: 2.0
-            gamma: 0.99
-        gail:
-            strength: 0.02
-            gamma: 0.99
-            encoding_size: 128
-            use_actions: true
-            demo_path: Project/Assets/ML-Agents/Examples/Pyramids/Demos/ExpertPyramid.demo
-
-VisualPyramids:
-    time_horizon: 128
-    batch_size: 64
-    hidden_units: 256
-    buffer_init_steps: 1000
-    num_layers: 1
-    max_steps: 1.0e7
-    buffer_size: 500000
-    init_entcoef: 0.01
-    tau: 0.01
-    reward_signals:
-        extrinsic:
-            strength: 2.0
-            gamma: 0.99
-        gail:
-            strength: 0.02
-            gamma: 0.99
-            encoding_size: 128
-            use_actions: true
-            demo_path: Project/Assets/ML-Agents/Examples/Pyramids/Demos/ExpertPyramid.demo
-
-3DBall:
-    normalize: true
-    batch_size: 64
-    buffer_size: 12000
-    summary_freq: 12000
-    time_horizon: 1000
-    hidden_units: 64
-    init_entcoef: 0.5
-
-3DBallHard:
-    normalize: true
-    batch_size: 256
-    summary_freq: 12000
-    time_horizon: 1000
-
-Tennis:
-    normalize: true
-    max_steps: 2e7
-    hidden_units: 256
-    self_play:
-        window: 10
-        play_against_current_self_ratio: 0.5
-        save_steps: 50000
-        swap_steps: 50000
-
-CrawlerStatic:
-    normalize: true
-    time_horizon: 1000
-    batch_size: 256
-    steps_per_update: 20
-    buffer_size: 500000
-    buffer_init_steps: 2000
-    max_steps: 3e6
-    summary_freq: 30000
-    init_entcoef: 1.0
-    num_layers: 3
-    hidden_units: 512
-    reward_signals:
-        extrinsic:
-            strength: 1.0
-            gamma: 0.995
-
-CrawlerDynamic:
-    normalize: true
-    time_horizon: 1000
-    batch_size: 256
-    buffer_size: 500000
-    summary_freq: 30000
-    steps_per_update: 20
-    num_layers: 3
-    max_steps: 5e6
-    hidden_units: 512
-    reward_signals:
-        extrinsic:
-            strength: 1.0
-            gamma: 0.995
-
-Walker:
-    normalize: true
-    time_horizon: 1000
-    batch_size: 256
-    buffer_size: 500000
-    max_steps: 2e7
-    summary_freq: 30000
-    num_layers: 4
-    steps_per_update: 30
-    hidden_units: 512
-    reward_signals:
-        extrinsic:
-            strength: 1.0
-            gamma: 0.995
-
-Reacher:
-    normalize: true
-    time_horizon: 1000
-    batch_size: 128
-    buffer_size: 500000
-    max_steps: 2e7
-    steps_per_update: 20
-    summary_freq: 60000
-
-Hallway:
-    sequence_length: 32
-    num_layers: 2
-    hidden_units: 128
-    memory_size: 128
-    init_entcoef: 0.1
-    max_steps: 5.0e6
-    summary_freq: 10000
-    time_horizon: 64
-    use_recurrent: true
-
-VisualHallway:
-    sequence_length: 32
-    num_layers: 1
-    hidden_units: 128
-    memory_size: 128
-    gamma: 0.99
-    batch_size: 64
-    max_steps: 1.0e7
-    summary_freq: 10000
-    time_horizon: 64
-    use_recurrent: true
-
-VisualPushBlock:
-    use_recurrent: true
-    sequence_length: 32
-    num_layers: 1
-    hidden_units: 128
-    memory_size: 128
-    gamma: 0.99
-    buffer_size: 1024
-    batch_size: 64
-    max_steps: 3.0e6
-    summary_freq: 60000
-    time_horizon: 64
-
-GridWorld:
-    batch_size: 128
-    normalize: false
-    num_layers: 1
-    hidden_units: 128
-    init_entcoef: 0.5
-    buffer_init_steps: 1000
-    buffer_size: 50000
-    max_steps: 500000
-    summary_freq: 20000
-    time_horizon: 5
-    reward_signals:
-        extrinsic:
-            strength: 1.0
-            gamma: 0.9
-
-Basic:
-    batch_size: 64
-    normalize: false
-    num_layers: 2
-    init_entcoef: 0.01
-    hidden_units: 20
-    max_steps: 5.0e5
-    summary_freq: 2000
-    time_horizon: 10
-
-WormStatic:
-    normalize: true
-    time_horizon: 1000
-    batch_size: 128
-    steps_per_update: 20
-    buffer_size: 500000
-    buffer_init_steps: 2000
-    max_steps: 3e6
-    summary_freq: 30000
-    init_entcoef: 1.0
-    num_layers: 2
-    hidden_units: 256
-    reward_signals:
-        extrinsic:
-            strength: 1.0
-            gamma: 0.995
-
-WormDynamic:
-    normalize: true
-    time_horizon: 1000
-    batch_size: 128
-    buffer_size: 500000
-    summary_freq: 30000
-    steps_per_update: 20
-    num_layers: 2
-    max_steps: 5e6
-    hidden_units: 256
-    reward_signals:
-        extrinsic:
-            strength: 1.0
-            gamma: 0.995