[Feature] SAC Trainer

ghstack-source-id: 2db05ee
Pull-Request: #3191

Author: vmoens

torchrl/trainers/__init__.py

@@ -9,6 +9,7 @@
     CountFramesLog,
     LogReward,
     LogScalar,
+    LogTiming,
     LogValidationReward,
     mask_batch,
     OptimizerHook,
@@ -29,6 +30,7 @@
     "CountFramesLog",
     "LogReward",
     "LogScalar",
+    "LogTiming",
     "LogValidationReward",
     "mask_batch",
     "OptimizerHook",

torchrl/trainers/algorithms/sac.py

@@ -125,6 +125,7 @@ def __init__(
         log_observations: bool = False,
         target_net_updater: TargetNetUpdater | None = None,
         async_collection: bool = False,
+        log_timings: bool = False,
     ) -> None:
         warnings.warn(
             "SACTrainer is an experimental/prototype feature. The API may change in future versions. "
@@ -151,6 +152,7 @@ def __init__(
             log_interval=log_interval,
             save_trainer_file=save_trainer_file,
             async_collection=async_collection,
+            log_timings=log_timings,
         )
         self.replay_buffer = replay_buffer
         self.async_collection = async_collection