chore: type hints

instadeepai · Oct 22, 2024 · e13a6e1 · e13a6e1
1 parent 3b8d761
commit e13a6e1
Show file tree

Hide file tree

Showing 3 changed files with 28 additions and 24 deletions.
diff --git a/mava/systems/q_learning/anakin/rec_iql.py b/mava/systems/q_learning/anakin/rec_iql.py
@@ -443,7 +443,9 @@ def update_q(
 
         return next_params, next_opt_state, q_loss_info
 
-    def train(train_state: TrainState, _: Any) -> Tuple[TrainState, Metrics]:
+    def train(
+        train_state: TrainState[QNetParams], _: Any
+    ) -> Tuple[TrainState[QNetParams], Metrics]:
         """Sample, train and repack."""
         # unpack and get keys
         buffer_state, params, opt_states, t_train, key = train_state

diff --git a/mava/systems/q_learning/anakin/rec_qmix.py b/mava/systems/q_learning/anakin/rec_qmix.py
@@ -28,7 +28,6 @@
 from flax.core.scope import FrozenVariableDict
 from flax.linen import FrozenDict
 from jax import Array, tree
-from jumanji.env import Environment
 from jumanji.types import TimeStep
 from omegaconf import DictConfig, OmegaConf
 from rich.pretty import pprint
@@ -46,7 +45,7 @@
     TrainState,
     Transition,
 )
-from mava.types import Observation, ObservationGlobalState
+from mava.types import MarlEnv, Observation
 from mava.utils import make_env as environments
 from mava.utils.checkpointing import Checkpointer
 from mava.utils.jax_utils import (
@@ -59,12 +58,14 @@
 from mava.wrappers import episode_metrics
 
 
+# (env, eval_env), learner_state, q_net, q_mixer, opt, rb, logger, key
 def init(
     cfg: DictConfig,
 ) -> Tuple[
-    Tuple[Environment, Environment],
+    Tuple[MarlEnv, MarlEnv],
     LearnerState,
     RecQNetwork,
+    QMixNetwork,
     optax.GradientTransformation,
     TrajectoryBuffer,
     MavaLogger,
@@ -222,15 +223,15 @@ def replicate(x: Any) -> Any:
 
 def make_update_fns(
     cfg: DictConfig,
-    env: Environment,
+    env: MarlEnv,
     q_net: RecQNetwork,
     mixer: QMixNetwork,
     opt: optax.GradientTransformation,
     rb: TrajectoryBuffer,
 ) -> Any:
     def select_eps_greedy_action(
         action_selection_state: ActionSelectionState,
-        obs: ObservationGlobalState,
+        obs: Observation,
         term_or_trunc: Array,
     ) -> Tuple[ActionSelectionState, Array]:
         """Select action to take in eps-greedy way. Batch and agent dims are included."""
@@ -364,12 +365,8 @@ def update_q(
         """Update the Q parameters."""
 
         # Get data aligned with current/next timestep
-        data_first: Dict[str, chex.Array] = jax.tree_map(
-            lambda x: x[:, :-1, ...], data
-        )  # (B, T, ...)
-        data_next: Dict[str, chex.Array] = jax.tree_map(
-            lambda x: x[:, 1:, ...], data
-        )  # (B, T, ...)
+        data_first = jax.tree_map(lambda x: x[:, :-1, ...], data)  # (B, T, ...)
+        data_next = jax.tree_map(lambda x: x[:, 1:, ...], data)  # (B, T, ...)
 
         first_reward = data_first.reward
         next_done = data_next.term_or_trunc
@@ -450,7 +447,9 @@ def update_q(
 
         return next_params, next_opt_state, q_loss_info
 
-    def train(train_state: TrainState, _: Any) -> TrainState:
+    def train(
+        train_state: TrainState[QMIXParams], _: Any
+    ) -> Tuple[TrainState[QMIXParams], Metrics]:
         """Sample, train and repack."""
 
         buffer_state, params, opt_states, t_train, key = train_state

diff --git a/mava/systems/q_learning/types.py b/mava/systems/q_learning/types.py
@@ -11,15 +11,15 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-from typing import Dict, NamedTuple
+from typing import Dict, Generic, TypeVar
 
 import optax
 from chex import PRNGKey
 from flashbax.buffers.trajectory_buffer import TrajectoryBufferState
 from flax.core.scope import FrozenVariableDict
 from jax import Array
 from jumanji.env import State
-from typing_extensions import TypeAlias
+from typing_extensions import NamedTuple, TypeAlias
 
 from mava.types import Observation
 
@@ -90,18 +90,21 @@ class ActionState(NamedTuple):
     term_or_trunc: Array
 
 
-class TrainState(NamedTuple):
+class QMIXParams(NamedTuple):
+    online: FrozenVariableDict
+    target: FrozenVariableDict
+    mixer_online: FrozenVariableDict
+    mixer_target: FrozenVariableDict
+
+
+QLearningParams = TypeVar("QLearningParams", QNetParams, QMIXParams)
+
+
+class TrainState(NamedTuple, Generic[QLearningParams]):
     """The carry in the training loop."""
 
     buffer_state: BufferState
-    params: QNetParams
+    params: QLearningParams
     opt_state: optax.OptState
     train_steps: Array
     key: PRNGKey
-
-
-class QMIXParams(NamedTuple):
-    online: FrozenVariableDict
-    target: FrozenVariableDict
-    mixer_online: FrozenVariableDict
-    mixer_target: FrozenVariableDict