Merge branch 'master' into asymm-envs

Project/Assets/ML-Agents/Examples/SharedAssets/Scripts/ModelOverrider.cs

 using System;
 using System.Collections.Generic;
 using UnityEngine;
-using Barracuda;
+using Unity.Barracuda;
 using System.IO;
 using MLAgents;
 using MLAgents.Policies;

Project/Assets/ML-Agents/Examples/WallJump/Scripts/WallJumpAgent.cs

 using System.Collections;
 using UnityEngine;
 using MLAgents;
-using Barracuda;
+using Unity.Barracuda;
 using MLAgents.Sensors;
 using MLAgents.SideChannels;
 

com.unity.ml-agents/.npmignore

 build.sh.meta
 build.bat
 build.bat.meta
+upm-ci.log

com.unity.ml-agents/CHANGELOG.md

 
 ## [Unreleased]
 ### Major Changes
+#### com.unity.ml-agents (C#)
+#### ml-agents / ml-agents-envs / gym-unity (Python)
+#### com.unity.ml-agents (C#)
+#### ml-agents / ml-agents-envs / gym-unity (Python)
+#### com.unity.ml-agents (C#)
+#### ml-agents / ml-agents-envs / gym-unity (Python)
 
 ## [1.0.0-preview] - 2020-05-06
 
   C# style conventions. All public fields and properties now use "PascalCase"
   instead of "camelCase"; for example, `Agent.maxStep` was renamed to
   `Agent.MaxStep`. For a full list of changes, see the pull request. (#3828)
+- Update Barracuda to 0.7.0-preview which has breaking namespace and assembly name changes.
 
 ### Minor Changes
 
 - `StackingSensor` was changed from `internal` visibility to `public`
 - Academy.InferenceSeed property was added. This is used to initialize the
   random number generator in ModelRunner, and is incremented for each ModelRunner. (#3823)
-- Updated Barracuda to 0.6.3-preview.
 - Added `Agent.GetObservations(), which returns a read-only view of the observations
   added in `CollectObservations()`. (#3825)
 - Model updates can now happen asynchronously with environment steps for better performance. (#3690)

com.unity.ml-agents/Editor/BehaviorParametersEditor.cs

 using MLAgents.Sensors;
 using UnityEditor;
-using Barracuda;
+using Unity.Barracuda;
 using MLAgents.Policies;
 using UnityEngine;
 

com.unity.ml-agents/Editor/Unity.ML-Agents.Editor.asmdef

     "name": "Unity.ML-Agents.Editor",
     "references": [
         "Unity.ML-Agents",
-        "Barracuda",
+        "Unity.Barracuda",
         "Unity.ML-Agents.CommunicatorObjects"
     ],
     "optionalUnityReferences": [],
     "precompiledReferences": [],
     "autoReferenced": true,
     "defineConstraints": []
-}
+}

com.unity.ml-agents/Runtime/Academy.cs

 using MLAgents.Inference;
 using MLAgents.Policies;
 using MLAgents.SideChannels;
-using Barracuda;
+using Unity.Barracuda;
 
 /**
  * Welcome to Unity Machine Learning Agents (ML-Agents).

com.unity.ml-agents/Runtime/Agent.cs

 using System.Collections.Generic;
 using System.Collections.ObjectModel;
 using UnityEngine;
-using Barracuda;
+using Unity.Barracuda;
 using MLAgents.Sensors;
 using MLAgents.Demonstrations;
 using MLAgents.Policies;

com.unity.ml-agents/Runtime/Inference/ApplierImpl.cs

 using System;
 using System.Collections.Generic;
 using System.Linq;
-using Barracuda;
+using Unity.Barracuda;
 using MLAgents.Inference.Utils;
 using UnityEngine;
 

com.unity.ml-agents/Runtime/Inference/BarracudaModelParamLoader.cs

 using System;
 using System.Collections.Generic;
 using System.Linq;
-using Barracuda;
+using Unity.Barracuda;
 using MLAgents.Sensors;
 using MLAgents.Policies;
 
                     "suggest Continuous Control.");
                 return failedModelChecks;
             }
-            var tensorTester = new Dictionary<string, Func<BrainParameters, TensorShape, int, string>>();
+            var tensorTester = new Dictionary<string, Func<BrainParameters, TensorShape?, int, string>>();
             if (brainParameters.VectorActionSpaceType == SpaceType.Continuous)
             {
                 tensorTester[TensorNames.ActionOutput] = CheckContinuousActionOutputShape;
             {
                 if (tensorTester.ContainsKey(name))
                 {
-                    var tester = tensorTester[name];
+                    Func<BrainParameters, TensorShape?, int, string> tester = tensorTester[name];
                     var error = tester.Invoke(brainParameters, model.GetShapeByName(name), modelActionSize);
                     if (error != null)
                     {
         /// check failed. If the check passed, returns null.
         /// </returns>
         static string CheckDiscreteActionOutputShape(
-            BrainParameters brainParameters, TensorShape shape, int modelActionSize)
+            BrainParameters brainParameters, TensorShape? shape, int modelActionSize)
         {
             var bpActionSize = brainParameters.VectorActionSize.Sum();
             if (modelActionSize != bpActionSize)
         /// <returns>If the Check failed, returns a string containing information about why the
         /// check failed. If the check passed, returns null.</returns>
         static string CheckContinuousActionOutputShape(
-            BrainParameters brainParameters, TensorShape shape, int modelActionSize)
+            BrainParameters brainParameters, TensorShape? shape, int modelActionSize)
         {
             var bpActionSize = brainParameters.VectorActionSize[0];
             if (modelActionSize != bpActionSize)

com.unity.ml-agents/Runtime/Inference/GeneratorImpl.cs

 using System.Collections.Generic;
 using System;
-using Barracuda;
+using Unity.Barracuda;
 using MLAgents.Inference.Utils;
 using UnityEngine;
 using MLAgents.Sensors;

com.unity.ml-agents/Runtime/Inference/ModelRunner.cs

 using System.Collections.Generic;
-using Barracuda;
+using Unity.Barracuda;
 using UnityEngine.Profiling;
 using MLAgents.Sensors;
 using MLAgents.Policies;

com.unity.ml-agents/Runtime/Inference/TensorApplier.cs

 using System.Collections.Generic;
-using Barracuda;
+using Unity.Barracuda;
 using MLAgents.Policies;
 
 namespace MLAgents.Inference

com.unity.ml-agents/Runtime/Inference/TensorGenerator.cs

 using System.Collections.Generic;
-using Barracuda;
+using Unity.Barracuda;
 using MLAgents.Sensors;
 
 namespace MLAgents.Inference

com.unity.ml-agents/Runtime/Inference/TensorProxy.cs

 using System;
 using System.Collections.Generic;
-using Barracuda;
+using Unity.Barracuda;
 using MLAgents.Inference.Utils;
 
 namespace MLAgents.Inference

com.unity.ml-agents/Runtime/Policies/BarracudaPolicy.cs

-using Barracuda;
+using Unity.Barracuda;
 using System.Collections.Generic;
 using MLAgents.Inference;
 using MLAgents.Sensors;

com.unity.ml-agents/Runtime/Policies/BehaviorParameters.cs

-using Barracuda;
+using Unity.Barracuda;
 using System;
 using UnityEngine;
 using UnityEngine.Serialization;

com.unity.ml-agents/Runtime/Sensors/ObservationWriter.cs

 using System;
 using System.Collections.Generic;
-using Barracuda;
+using Unity.Barracuda;
 using MLAgents.Inference;
 
 namespace MLAgents.Sensors

com.unity.ml-agents/Runtime/Unity.ML-Agents.asmdef

 {
     "name": "Unity.ML-Agents",
     "references": [
-        "Barracuda",
+        "Unity.Barracuda",
         "Unity.ML-Agents.CommunicatorObjects"
     ],
     "optionalUnityReferences": [],
     ],
     "autoReferenced": true,
     "defineConstraints": []
-}
+}

com.unity.ml-agents/Tests/Editor/DiscreteActionOutputApplierTest.cs

 using System;
-using Barracuda;
+using Unity.Barracuda;
 using NUnit.Framework;
 using UnityEngine;
 using MLAgents.Inference;

com.unity.ml-agents/Tests/Editor/EditModeTestInternalBrainTensorApplier.cs

 using System.Collections.Generic;
 using NUnit.Framework;
-using Barracuda;
+using Unity.Barracuda;
 using MLAgents.Inference;
 using MLAgents.Policies;
 

com.unity.ml-agents/Tests/Editor/EditModeTestInternalBrainTensorGenerator.cs

 using System.Collections.Generic;
-using Barracuda;
+using Unity.Barracuda;
 using NUnit.Framework;
 using UnityEngine;
 using MLAgents.Inference;

com.unity.ml-agents/Tests/Editor/ModelRunnerTest.cs

 using NUnit.Framework;
 using UnityEngine;
 using UnityEditor;
-using Barracuda;
+using Unity.Barracuda;
 using MLAgents.Inference;
 using MLAgents.Sensors;
 using System.Linq;

com.unity.ml-agents/Tests/Editor/ParameterLoaderTest.cs

 using NUnit.Framework;
 using UnityEngine;
 using UnityEditor;
-using Barracuda;
+using Unity.Barracuda;
 using MLAgents.Inference;
 using MLAgents.Sensors;
 using System.Linq;

com.unity.ml-agents/Tests/Editor/Sensor/ObservationWriterTests.cs

 using NUnit.Framework;
 using MLAgents.Sensors;
-using Barracuda;
+using Unity.Barracuda;
 using MLAgents.Inference;
 
 

com.unity.ml-agents/Tests/Editor/TensorUtilsTest.cs

 using System;
-using Barracuda;
+using Unity.Barracuda;
 using MLAgents.Inference;
 using MLAgents.Inference.Utils;
 using NUnit.Framework;

com.unity.ml-agents/Tests/Editor/Unity.ML-Agents.Editor.Tests.asmdef

     "references": [
         "Unity.ML-Agents.Editor",
         "Unity.ML-Agents",
-        "Barracuda",
+        "Unity.Barracuda",
         "Unity.ML-Agents.CommunicatorObjects"
     ],
     "optionalUnityReferences": [
     "defineConstraints": [
         "UNITY_INCLUDE_TESTS"
     ]
-}
+}

com.unity.ml-agents/Tests/Runtime/Unity.ML-Agents.Runtime.Tests.asmdef

     "name": "Tests",
     "references": [
         "Unity.ML-Agents",
-        "Barracuda",
+        "Unity.Barracuda",
         "Unity.ML-Agents.CommunicatorObjects",
         "Unity.ML-Agents.Editor"
     ],

com.unity.ml-agents/package.json

   "unity": "2018.4",
   "description": "Add interactivity to your game with Machine Learning Agents trained using Deep Reinforcement Learning.",
   "dependencies": {
-    "com.unity.barracuda": "0.6.3-preview"
+    "com.unity.barracuda": "0.7.0-preview"
-}
+}

ml-agents/mlagents/trainers/ghost/controller.py

         self._learning_team: int = -1
         # Dict from team id to GhostTrainer for ELO calculation
         self._ghost_trainers: Dict[int, GhostTrainer] = {}
-        # Signals to the trainer control to perform a hard reset
-        self._reset = False
+        # Signals to the trainer control to perform a hard change_training_team
+        self._changed_training_team = False
 
     @property
     def get_learning_team(self) -> int:
         """
         return self._learning_team
 
-    @property
-    def reset(self) -> bool:
+    def should_reset(self) -> bool:
-        change_team = self._reset
-        if self._reset:
-            self._reset = False
-        return change_team
+        changed_team = self._changed_training_team
+        if self._changed_training_team:
+            self._changed_training_team = False
+        return changed_team
 
     def subscribe_team_id(self, team_id: int, trainer: GhostTrainer) -> None:
         """
         logger.debug(
             "Learning team {} swapped on step {}".format(self._learning_team, step)
         )
-        self._reset = True
+        self._changed_training_team = True
 
     # Adapted from https://github.com/Unity-Technologies/ml-agents/pull/1975 and
     # https://metinmediamath.wordpress.com/2013/11/27/how-to-calculate-the-elo-rating-including-example/

ml-agents/mlagents/trainers/tests/simple_test_envs.py

         self.goal[name] = self.random.choice([-1, 1])
         self.positions[name] = [0.0 for _ in range(self.action_size)]
         self.step_count[name] = 0
-        self.final_rewards[name].append(self.rewards[name])
         self.rewards[name] = 0
         self.agent_id[name] = self.agent_id[name] + 1
 
         decision_step = DecisionSteps(m_vector_obs, m_reward, m_agent_id, action_mask)
         terminal_step = TerminalSteps.empty(self.behavior_spec)
         if done:
+            self.final_rewards[name].append(self.rewards[name])
             self._reset_agent(name)
             new_vector_obs = self._make_obs(self.goal[name])
             (
         decision_step = DecisionSteps(m_vector_obs, m_reward, m_agent_id, action_mask)
         terminal_step = TerminalSteps.empty(self.behavior_spec)
         if done:
+            self.final_rewards[name].append(self.rewards[name])
             self._reset_agent(name)
             recurrent_obs_val = (
                 self.goal[name] if self.step_count[name] <= self.num_show_steps else 0

ml-agents/mlagents/trainers/tests/test_trainer_controller.py

 
 from mlagents.tf_utils import tf
 from mlagents.trainers.trainer_controller import TrainerController
+from mlagents.trainers.ghost.controller import GhostController
+    trainer_factory_mock = MagicMock()
+    trainer_factory_mock.ghost_controller = GhostController()
-        trainer_factory=None,
+        trainer_factory=trainer_factory_mock,
         model_path="test_model_path",
         summaries_dir="test_summaries_dir",
         run_id="test_run_id",
 @patch.object(tf, "set_random_seed")
 def test_initialization_seed(numpy_random_seed, tensorflow_set_seed):
     seed = 27
+    trainer_factory_mock = MagicMock()
+    trainer_factory_mock.ghost_controller = GhostController()
-        trainer_factory=None,
+        trainer_factory=trainer_factory_mock,
         model_path="",
         summaries_dir="",
         run_id="1",

ml-agents/mlagents/trainers/trainer_controller.py

         self.meta_curriculum = meta_curriculum
         self.sampler_manager = sampler_manager
         self.resampling_interval = resampling_interval
-        self.ghost_controller = ghost_controller
+        self.ghost_controller = self.trainer_factory.ghost_controller
 
         self.trainer_threads: List[threading.Thread] = []
         self.kill_trainers = False
             and (self.resampling_interval)
             and (steps % self.resampling_interval == 0)
         )
-        if meta_curriculum_reset or generalization_reset or self.ghost_controller.reset:
+        ghost_controller_reset = self.ghost_controller.should_reset()
+        if meta_curriculum_reset or generalization_reset or ghost_controller_reset:
             self.end_trainer_episodes(env, lessons_incremented)
 
     @timed