Make sure all tests pass on BC

Author: Ervin Teng

ml-agents/mlagents/trainers/bc/policy.py

@@ -79,24 +79,14 @@ def update(self, mini_batch, num_sequences):
             self.model.sequence_length: self.sequence_length,
         }
         if self.use_continuous_act:
-            feed_dict[self.model.true_action] = mini_batch["actions"].reshape(
-                [-1, self.brain.vector_action_space_size[0]]
-            )
+            feed_dict[self.model.true_action] = mini_batch["actions"]
         else:
-            feed_dict[self.model.true_action] = mini_batch["actions"].reshape(
-                [-1, len(self.brain.vector_action_space_size)]
-            )
+            feed_dict[self.model.true_action] = mini_batch["actions"]
             feed_dict[self.model.action_masks] = np.ones(
                 (num_sequences, sum(self.brain.vector_action_space_size))
             )
         if self.use_vec_obs:
-            apparent_obs_size = (
-                self.brain.vector_observation_space_size
-                * self.brain.num_stacked_vector_observations
-            )
-            feed_dict[self.model.vector_in] = mini_batch["vector_obs"].reshape(
-                [-1, apparent_obs_size]
-            )
+            feed_dict[self.model.vector_in] = mini_batch["vector_obs"]
         for i, _ in enumerate(self.model.visual_in):
             visual_obs = mini_batch["visual_obs%d" % i]
             feed_dict[self.model.visual_in[i]] = visual_obs

ml-agents/mlagents/trainers/bc/trainer.py

@@ -124,7 +124,7 @@ def update_policy(self):
         """
         Updates the policy.
         """
-        self.demonstration_buffer.update_buffer.shuffle()
+        self.demonstration_buffer.update_buffer.shuffle(self.policy.sequence_length)
         batch_losses = []
         num_batches = min(
             len(self.demonstration_buffer.update_buffer["actions"]) // self.n_sequences,