[RLlib] Add support for custom MultiActionDistributions. (#11311)

rllib/models/catalog.py

@@ -3,7 +3,7 @@ import gym
 import logging
 import numpy as np
 import tree
-from typing import List
+from typing import List, Optional, Type, Union
 
 from ray.tune.registry import RLLIB_MODEL, RLLIB_PREPROCESSOR, \
     RLLIB_ACTION_DIST, _global_registry
@@ -110,18 +110,20 @@ class ModelCatalog:
 
     @staticmethod
     @DeveloperAPI
-    def get_action_dist(action_space: gym.Space,
-                        config: ModelConfigDict,
-                        dist_type: str = None,
-                        framework: str = "tf",
-                        **kwargs) -> (type, int):
+    def get_action_dist(
+            action_space: gym.Space,
+            config: ModelConfigDict,
+            dist_type: Optional[Union[str, Type[ActionDistribution]]] = None,
+            framework: str = "tf",
+            **kwargs) -> (type, int):
         """Returns a distribution class and size for the given action space.
 
         Args:
             action_space (Space): Action space of the target gym env.
             config (Optional[dict]): Optional model config.
-            dist_type (Optional[str]): Identifier of the action distribution
-                type (str) interpreted as a hint.
+            dist_type (Optional[Union[str, Type[ActionDistribution]]]):
+                Identifier of the action distribution (str) interpreted as a
+                hint or the actual ActionDistribution class to use.
             framework (str): One of "tf", "tfe", or "torch".
             kwargs (dict): Optional kwargs to pass on to the Distribution's
                 constructor.
@@ -143,6 +145,9 @@ class ModelCatalog:
                 "Using custom action distribution {}".format(action_dist_name))
             dist_cls = _global_registry.get(RLLIB_ACTION_DIST,
                                             action_dist_name)
+            dist_cls = ModelCatalog._get_multi_action_distribution(
+                dist_cls, action_space, {}, framework)
+
         # Dist_type is given directly as a class.
         elif type(dist_type) is type and \
                 issubclass(dist_type, ActionDistribution) and \
@@ -173,18 +178,10 @@ class ModelCatalog:
         elif dist_type in (MultiActionDistribution,
                            TorchMultiActionDistribution) or \
                 isinstance(action_space, (gym.spaces.Tuple, gym.spaces.Dict)):
-            flat_action_space = flatten_space(action_space)
-            child_dists_and_in_lens = tree.map_structure(
-                lambda s: ModelCatalog.get_action_dist(
-                    s, config, framework=framework), flat_action_space)
-            child_dists = [e[0] for e in child_dists_and_in_lens]
-            input_lens = [int(e[1]) for e in child_dists_and_in_lens]
-            return partial(
-                (TorchMultiActionDistribution
-                 if framework == "torch" else MultiActionDistribution),
-                action_space=action_space,
-                child_distributions=child_dists,
-                input_lens=input_lens), int(sum(input_lens))
+            return ModelCatalog._get_multi_action_distribution(
+                (MultiActionDistribution
+                 if framework == "tf" else TorchMultiActionDistribution),
+                action_space, config, framework)
         # Simplex -> Dirichlet.
         elif isinstance(action_space, Simplex):
             if framework == "torch":
@@ -422,7 +419,8 @@ class ModelCatalog:
 
     @staticmethod
     @DeveloperAPI
-    def get_preprocessor(env: gym.Env, options: dict = None) -> Preprocessor:
+    def get_preprocessor(env: gym.Env,
+                         options: Optional[dict] = None) -> Preprocessor:
         """Returns a suitable preprocessor for the given env.
 
         This is a wrapper for get_preprocessor_for_space().
@@ -552,3 +550,25 @@ class ModelCatalog:
         # Default Conv2D net.
         else:
             return VisionNet
+
+    @staticmethod
+    def _get_multi_action_distribution(dist_class, action_space, config,
+                                       framework):
+        # In case the custom distribution is a child of MultiActionDistr.
+        # If users want to completely ignore the suggested child
+        # distributions, they should simply do so in their custom class'
+        # constructor.
+        if issubclass(dist_class,
+                      (MultiActionDistribution, TorchMultiActionDistribution)):
+            flat_action_space = flatten_space(action_space)
+            child_dists_and_in_lens = tree.map_structure(
+                lambda s: ModelCatalog.get_action_dist(
+                    s, config, framework=framework), flat_action_space)
+            child_dists = [e[0] for e in child_dists_and_in_lens]
+            input_lens = [int(e[1]) for e in child_dists_and_in_lens]
+            return partial(
+                dist_class,
+                action_space=action_space,
+                child_distributions=child_dists,
+                input_lens=input_lens), int(sum(input_lens))
+        return dist_class

rllib/tests/test_catalog.py

@@ -61,7 +61,7 @@ class CustomActionDistribution(TFActionDistribution):
         return tf.zeros(self.output_shape)
 
 
-class ModelCatalogTest(unittest.TestCase):
+class TestModelCatalog(unittest.TestCase):
     def tearDown(self):
         ray.shutdown()