[RLlib] Rename all np.product usage to np.prod (#46317)

python/ray/data/_internal/datasource/range_datasource.py

@@ -28,7 +28,7 @@ def __init__(
 
     def estimate_inmemory_data_size(self) -> Optional[int]:
         if self._block_format == "tensor":
-            element_size = int(np.product(self._tensor_shape))
+            element_size = int(np.prod(self._tensor_shape))
         else:
             element_size = 1
         return 8 * self._n * element_size
@@ -86,7 +86,7 @@ def make_blocks(
                 count -= num_rows
 
         if block_format == "tensor":
-            element_size = int(np.product(tensor_shape))
+            element_size = int(np.prod(tensor_shape))
         else:
             element_size = 1
 

rllib/algorithms/sac/rnnsac_torch_policy.py

@@ -43,7 +43,7 @@ def build_rnnsac_model(
             `policy.target_model`.
     """
     # With separate state-preprocessor (before obs+action concat).
-    num_outputs = int(np.product(obs_space.shape))
+    num_outputs = int(np.prod(obs_space.shape))
 
     # Force-ignore any additionally provided hidden layer sizes.
     # Everything should be configured using SAC's `q_model_config` and

rllib/algorithms/sac/sac_tf_model.py

@@ -77,13 +77,13 @@ def __init__(
             self.discrete = True
             action_outs = q_outs = self.action_dim
         elif isinstance(action_space, Box):
-            self.action_dim = np.product(action_space.shape)
+            self.action_dim = np.prod(action_space.shape)
             self.discrete = False
             action_outs = 2 * self.action_dim
             q_outs = 1
         else:
             assert isinstance(action_space, Simplex)
-            self.action_dim = np.product(action_space.shape)
+            self.action_dim = np.prod(action_space.shape)
             self.discrete = False
             action_outs = self.action_dim
             q_outs = 1

rllib/algorithms/sac/sac_torch_model.py

@@ -79,13 +79,13 @@ def __init__(
             self.discrete = True
             action_outs = q_outs = self.action_dim
         elif isinstance(action_space, Box):
-            self.action_dim = np.product(action_space.shape)
+            self.action_dim = np.prod(action_space.shape)
             self.discrete = False
             action_outs = 2 * self.action_dim
             q_outs = 1
         else:
             assert isinstance(action_space, Simplex)
-            self.action_dim = np.product(action_space.shape)
+            self.action_dim = np.prod(action_space.shape)
             self.discrete = False
             action_outs = self.action_dim
             q_outs = 1

rllib/examples/_old_api_stack/models/batch_norm_model.py

@@ -184,7 +184,7 @@ def __init__(
         )
         nn.Module.__init__(self)
         layers = []
-        prev_layer_size = int(np.product(obs_space.shape))
+        prev_layer_size = int(np.prod(obs_space.shape))
         self._logits = None
 
         # Create layers 0 to second-last.

rllib/examples/_old_api_stack/models/shared_weights_model.py

@@ -171,7 +171,7 @@ def __init__(
 
         # Non-shared initial layer.
         self.first_layer = SlimFC(
-            int(np.product(observation_space.shape)),
+            int(np.prod(observation_space.shape)),
             64,
             activation_fn=nn.ReLU,
             initializer=torch.nn.init.xavier_uniform_,

rllib/models/catalog.py

@@ -385,7 +385,7 @@ def get_action_dist(
                 dist_cls = (
                     TorchMultiCategorical if framework == "torch" else MultiCategorical
                 )
-                num_cats = int(np.product(action_space.shape))
+                num_cats = int(np.prod(action_space.shape))
                 return (
                     partial(
                         dist_cls,
@@ -495,7 +495,7 @@ def get_action_shape(
                     size += 1
                 else:
                     all_discrete = False
-                    size += np.product(flat_action_space[i].shape)
+                    size += np.prod(flat_action_space[i].shape)
             size = int(size)
             return dl_lib.int32 if all_discrete else dl_lib.float32, (None, size)
         else:

rllib/models/preprocessors.py

@@ -38,7 +38,7 @@ def __init__(self, obs_space: gym.Space, options: dict = None):
         else:
             self._options = options
         self.shape = self._init_shape(obs_space, self._options)
-        self._size = int(np.product(self.shape))
+        self._size = int(np.prod(self.shape))
         self._i = 0
         self._obs_for_type_matching = self._obs_space.sample()
 
@@ -276,7 +276,7 @@ def _init_shape(self, obs_space: gym.Space, options: dict) -> List[int]:
                 size += preprocessor.size
             else:
                 preprocessor = None
-                size += int(np.product(space.shape))
+                size += int(np.prod(space.shape))
             self.preprocessors.append(preprocessor)
         return (size,)
 
@@ -315,7 +315,7 @@ def _init_shape(self, obs_space: gym.Space, options: dict) -> List[int]:
                 size += preprocessor.size
             else:
                 preprocessor = None
-                size += int(np.product(space.shape))
+                size += int(np.prod(space.shape))
             self.preprocessors.append(preprocessor)
         return (size,)
 

rllib/models/tf/attention_net.py

@@ -409,7 +409,7 @@ def __init__(
             elif isinstance(space, MultiDiscrete):
                 self.action_dim += np.sum(space.nvec)
             elif space.shape is not None:
-                self.action_dim += int(np.product(space.shape))
+                self.action_dim += int(np.prod(space.shape))
             else:
                 self.action_dim += int(len(space))
 

rllib/models/tf/complex_input_net.py

@@ -98,7 +98,7 @@ def __init__(self, obs_space, action_space, num_outputs, model_config, name):
                 concat_size += int(self.one_hot[i].num_outputs)
             # Everything else (1D Box).
             else:
-                size = int(np.product(component.shape))
+                size = int(np.prod(component.shape))
                 config = {
                     "fcnet_hiddens": model_config["fcnet_hiddens"],
                     "fcnet_activation": model_config.get("fcnet_activation"),

rllib/models/tf/fcnet.py

@@ -53,7 +53,7 @@ def __init__(
 
         # We are using obs_flat, so take the flattened shape as input.
         inputs = tf.keras.layers.Input(
-            shape=(int(np.product(obs_space.shape)),), name="observations"
+            shape=(int(np.prod(obs_space.shape)),), name="observations"
         )
         # Last hidden layer output (before logits outputs).
         last_layer = inputs
@@ -99,9 +99,7 @@ def __init__(
                 )(last_layer)
             # Adjust num_outputs to be the number of nodes in the last layer.
             else:
-                self.num_outputs = ([int(np.product(obs_space.shape))] + hiddens[-1:])[
-                    -1
-                ]
+                self.num_outputs = ([int(np.prod(obs_space.shape))] + hiddens[-1:])[-1]
 
         # Concat the log std vars to the end of the state-dependent means.
         if free_log_std and logits_out is not None:

rllib/models/tf/recurrent_net.py

@@ -149,7 +149,7 @@ def __init__(
         # is the input size for the LSTM layer.
         # If None, set it to the observation space.
         if self.num_outputs is None:
-            self.num_outputs = int(np.product(self.obs_space.shape))
+            self.num_outputs = int(np.prod(self.obs_space.shape))
 
         self.cell_size = model_config["lstm_cell_size"]
         self.use_prev_action = model_config["lstm_use_prev_action"]
@@ -164,7 +164,7 @@ def __init__(
             elif isinstance(space, MultiDiscrete):
                 self.action_dim += np.sum(space.nvec)
             elif space.shape is not None:
-                self.action_dim += int(np.product(space.shape))
+                self.action_dim += int(np.prod(space.shape))
             else:
                 self.action_dim += int(len(space))
 

rllib/models/torch/attention_net.py

@@ -289,7 +289,7 @@ def __init__(
             elif isinstance(space, MultiDiscrete):
                 self.action_dim += np.sum(space.nvec)
             elif space.shape is not None:
-                self.action_dim += int(np.product(space.shape))
+                self.action_dim += int(np.prod(space.shape))
             else:
                 self.action_dim += int(len(space))
 

rllib/models/torch/complex_input_net.py

@@ -118,7 +118,7 @@ def __init__(self, obs_space, action_space, num_outputs, model_config, name):
                 self.add_module("one_hot_{}".format(i), self.one_hot[i])
             # Everything else (1D Box).
             else:
-                size = int(np.product(component.shape))
+                size = int(np.prod(component.shape))
                 config = {
                     "fcnet_hiddens": model_config["fcnet_hiddens"],
                     "fcnet_activation": model_config.get("fcnet_activation"),

rllib/models/torch/fcnet.py

@@ -49,7 +49,7 @@ def __init__(
             num_outputs = num_outputs // 2
 
         layers = []
-        prev_layer_size = int(np.product(obs_space.shape))
+        prev_layer_size = int(np.prod(obs_space.shape))
         self._logits = None
 
         # Create layers 0 to second-last.
@@ -97,9 +97,7 @@ def __init__(
                     activation_fn=None,
                 )
             else:
-                self.num_outputs = ([int(np.product(obs_space.shape))] + hiddens[-1:])[
-                    -1
-                ]
+                self.num_outputs = ([int(np.prod(obs_space.shape))] + hiddens[-1:])[-1]
 
         # Layer to add the log std vars to the state-dependent means.
         if self.free_log_std and self._logits:
@@ -110,7 +108,7 @@ def __init__(
         self._value_branch_separate = None
         if not self.vf_share_layers:
             # Build a parallel set of hidden layers for the value net.
-            prev_vf_layer_size = int(np.product(obs_space.shape))
+            prev_vf_layer_size = int(np.prod(obs_space.shape))
             vf_layers = []
             for size in hiddens:
                 vf_layers.append(

rllib/models/torch/recurrent_net.py

@@ -145,7 +145,7 @@ def __init__(
         # is the input size for the LSTM layer.
         # If None, set it to the observation space.
         if self.num_outputs is None:
-            self.num_outputs = int(np.product(self.obs_space.shape))
+            self.num_outputs = int(np.prod(self.obs_space.shape))
 
         self.cell_size = model_config["lstm_cell_size"]
         self.time_major = model_config.get("_time_major", False)
@@ -161,7 +161,7 @@ def __init__(
             elif isinstance(space, MultiDiscrete):
                 self.action_dim += np.sum(space.nvec)
             elif space.shape is not None:
-                self.action_dim += int(np.product(space.shape))
+                self.action_dim += int(np.prod(space.shape))
             else:
                 self.action_dim += int(len(space))
 

rllib_contrib/a2c/pyproject.toml

@@ -15,4 +15,4 @@ requires-python = ">=3.7, <3.11"
 dependencies = ["gym[accept-rom-license]", "gymnasium[accept-rom-license, atari]==0.26.3", "ray[rllib]==2.5.0"]
 
 [project.optional-dependencies]
-    development = ["pytest>=7.2.2", "pre-commit==2.21.0", "tensorflow==2.11.0", "tensorflow-probability==0.19.0", "torch==1.12.0"]
+    development = ["pytest>=7.2.2", "pre-commit==2.21.0", "tensorflow==2.11.0", "tensorflow-probability==0.19.0", "torch==1.12.0", "numpy<2"]

rllib_contrib/a2c/requirements.txt

@@ -1,3 +1,4 @@
 tensorflow==2.11.1
 tensorflow-probability==0.19.0
 torch==1.12.0
+numpy<2

rllib_contrib/a3c/pyproject.toml

@@ -15,4 +15,4 @@ requires-python = ">=3.7, <3.11"
 dependencies = ["gym[accept-rom-license]", "gymnasium[mujoco]==0.26.3", "ray[rllib]==2.3.1"]
 
 [project.optional-dependencies]
-development = ["pytest>=7.2.2", "pre-commit==2.21.0", "tensorflow==2.11.0", "tensorflow-probability==0.19.0", "torch==1.12.0"]
+development = ["pytest>=7.2.2", "pre-commit==2.21.0", "tensorflow==2.11.0", "tensorflow-probability==0.19.0", "torch==1.12.0", "numpy<2"]

rllib_contrib/a3c/requirements.txt

@@ -1,3 +1,4 @@
 tensorflow==2.11.1
 tensorflow-probability==0.19.0
 torch==1.12.0
+numpy<2

rllib_contrib/alpha_star/pyproject.toml

@@ -15,4 +15,4 @@ requires-python = ">=3.7, <3.11"
 dependencies = ["gym", "gymnasium==0.26.3", "ray[rllib]==2.5.0", "open-spiel==1.3"]
 
 [project.optional-dependencies]
-development = ["pytest>=7.2.2", "pre-commit==2.21.0", "tensorflow==2.11.0", "tensorflow-probability==0.19.0", "torch==1.12.0"]
+development = ["pytest>=7.2.2", "pre-commit==2.21.0", "tensorflow==2.11.0", "tensorflow-probability==0.19.0", "torch==1.12.0", "numpy<2"]

rllib_contrib/alpha_star/requirements.txt

@@ -1,3 +1,4 @@
 tensorflow==2.11.1
 tensorflow-probability==0.19.0
 torch==1.12.0
+numpy<2

rllib_contrib/alpha_zero/pyproject.toml

@@ -15,4 +15,4 @@ requires-python = ">=3.7, <3.11"
 dependencies = ["gymnasium==0.26.3", "ray[rllib]==2.5.1"]
 
 [project.optional-dependencies]
-development = ["pytest>=7.2.2", "pre-commit==2.21.0", "tensorflow==2.11.0", "tensorflow-probability==0.19.0", "torch==1.12.0"]
+development = ["pytest>=7.2.2", "pre-commit==2.21.0", "tensorflow==2.11.0", "tensorflow-probability==0.19.0", "torch==1.12.0", "numpy<2"]

rllib_contrib/alpha_zero/requirements.txt

@@ -1 +1,2 @@
 torch==1.12.0
+numpy<2

rllib_contrib/apex_ddpg/pyproject.toml

@@ -15,4 +15,4 @@ requires-python = ">=3.7, <3.11"
 dependencies = ["gymnasium[atari]", "ray[rllib]==2.5.0"]
 
 [project.optional-dependencies]
-development = ["pytest>=7.2.2", "pre-commit==2.21.0", "tensorflow==2.11.0", "torch==1.12.0"]
+development = ["pytest>=7.2.2", "pre-commit==2.21.0", "tensorflow==2.11.0", "torch==1.12.0", "numpy<2"]

rllib_contrib/apex_ddpg/requirements.txt

@@ -1,3 +1,4 @@
 tensorflow==2.11.1
 tensorflow-probability==0.19.0
 torch==1.12.0
+numpy<2

rllib_contrib/apex_dqn/pyproject.toml

@@ -15,4 +15,4 @@ requires-python = ">=3.7, <3.11"
 dependencies = ["gymnasium[atari]", "ray[rllib]==2.5.0"]
 
 [project.optional-dependencies]
-development = ["pytest>=7.2.2", "pre-commit==2.21.0", "tensorflow==2.11.0", "torch==1.12.0"]
+development = ["pytest>=7.2.2", "pre-commit==2.21.0", "tensorflow==2.11.0", "torch==1.12.0", "numpy<2"]

rllib_contrib/apex_dqn/requirements.txt

@@ -1,3 +1,4 @@
 tensorflow==2.11.1
 tensorflow-probability==0.19.0
 torch==1.12.0
+numpy<2

rllib_contrib/ars/pyproject.toml

@@ -15,4 +15,4 @@ requires-python = ">=3.7, <3.11"
 dependencies = ["gymnasium", "ray[rllib]==2.5.0"]
 
 [project.optional-dependencies]
-development = ["pytest>=7.2.2", "pre-commit==2.21.0", "tensorflow==2.11.0", "tensorflow-probability==0.19.0", "torch==1.12.0"]
+development = ["pytest>=7.2.2", "pre-commit==2.21.0", "tensorflow==2.11.0", "tensorflow-probability==0.19.0", "torch==1.12.0", "numpy<2"]

rllib_contrib/ars/requirements.txt

@@ -1,3 +1,4 @@
 tensorflow==2.11.1
 tensorflow-probability==0.19.0
 torch==1.12.0
+numpy<2

rllib_contrib/crr/pyproject.toml

@@ -15,4 +15,4 @@ requires-python = ">=3.7, <3.11"
 dependencies = ["gymnasium", "ray[rllib]==2.5.0"]
 
 [project.optional-dependencies]
-development = ["pytest>=7.2.2", "pre-commit==2.21.0", "tensorflow==2.11.0", "tensorflow-probability==0.19.0", "torch==1.12.0"]
+development = ["pytest>=7.2.2", "pre-commit==2.21.0", "tensorflow==2.11.0", "tensorflow-probability==0.19.0", "torch==1.12.0", "numpy<2"]

rllib_contrib/crr/requirements.txt

@@ -1 +1,2 @@
 torch==1.12.0
+numpy<2

rllib_contrib/crr/src/rllib_crr/crr/crr_torch_model.py

@@ -30,7 +30,7 @@ def __init__(
 
         # TODO: I don't know why this is true yet? (in = num_outputs)
         self.obs_ins = num_outputs
-        self.action_dim = np.product(self.action_space.shape)
+        self.action_dim = np.prod(self.action_space.shape)
         self.actor_model = self._build_actor_net("actor")
         twin_q = self.model_config["twin_q"]
         self.q_model = self._build_q_net("q")

rllib_contrib/crr/src/rllib_crr/crr/crr_torch_policy.py

@@ -96,7 +96,7 @@ def make_model(self) -> ModelV2:
                 twin_q=self.config["twin_q"],
             )
         )
-        num_outputs = int(np.product(self.observation_space.shape))
+        num_outputs = int(np.prod(self.observation_space.shape))
 
         # TODO: why do we even have to go through this get_model_v2 function?
         self.model = ModelCatalog.get_model_v2(

rllib_contrib/ddpg/src/rllib_ddpg/ddpg/ddpg_torch_model.py

@@ -67,7 +67,7 @@ def __init__(
         self.bounded = np.logical_and(
             self.action_space.bounded_above, self.action_space.bounded_below
         ).any()
-        self.action_dim = np.product(self.action_space.shape)
+        self.action_dim = np.prod(self.action_space.shape)
 
         # Build the policy network.
         self.policy_model = nn.Sequential()

rllib_contrib/ddpg/src/rllib_ddpg/ddpg/utils.py

@@ -19,7 +19,7 @@ def make_ddpg_models(policy: Policy) -> ModelV2:
         default_model = (
             TorchNoopModel if policy.config["framework"] == "torch" else NoopModel
         )
-        num_outputs = int(np.product(policy.observation_space.shape))
+        num_outputs = int(np.prod(policy.observation_space.shape))
 
     model = ModelCatalog.get_model_v2(
         obs_space=policy.observation_space,

rllib_contrib/ddppo/pyproject.toml

@@ -15,4 +15,4 @@ requires-python = ">=3.7, <3.11"
 dependencies = ["gymnasium", "ray[rllib]==2.5.0"]
 
 [project.optional-dependencies]
-development = ["pytest>=7.2.2", "pre-commit==2.21.0", "torch==1.12.0", "tensorflow==2.11.0", "tensorflow-probability==0.19.0"]
+development = ["pytest>=7.2.2", "pre-commit==2.21.0", "torch==1.12.0", "tensorflow==2.11.0", "tensorflow-probability==0.19.0", "numpy<2"]

rllib_contrib/ddppo/requirements.txt

@@ -1 +1,2 @@
 torch==1.12.0
+numpy<2

rllib_contrib/dt/pyproject.toml

@@ -15,4 +15,4 @@ requires-python = ">=3.7, <3.11"
 dependencies = ["gymnasium", "ray[rllib]==2.5.0"]
 
 [project.optional-dependencies]
-development = ["pytest>=7.2.2", "pre-commit==2.21.0", "torch==1.12.0", "tensorflow==2.11.0", "tensorflow-probability==0.19.0"]
+development = ["pytest>=7.2.2", "pre-commit==2.21.0", "torch==1.12.0", "tensorflow==2.11.0", "tensorflow-probability==0.19.0", "numpy<2"]

rllib_contrib/dt/requirements.txt

@@ -1 +1,2 @@
 torch==1.12.0
+numpy<2

rllib_contrib/dt/src/rllib_dt/dt/dt_torch_model.py

@@ -35,7 +35,7 @@ def __init__(
         if isinstance(action_space, Discrete):
             self.action_dim = action_space.n
         elif isinstance(action_space, Box):
-            self.action_dim = np.product(action_space.shape)
+            self.action_dim = np.prod(action_space.shape)
         else:
             raise NotImplementedError
 

rllib_contrib/dt/src/rllib_dt/dt/dt_torch_policy.py

@@ -76,7 +76,7 @@ def make_model_and_action_dist(
             use_return_output=self.config.get("loss_coef_returns_to_go", 0) > 0,
         )
 
-        num_outputs = int(np.product(self.observation_space.shape))
+        num_outputs = int(np.prod(self.observation_space.shape))
 
         model = ModelCatalog.get_model_v2(
             obs_space=self.observation_space,