Makes automatic optimization a model attribute (#4602)

* Makes automatic optimization a model attribute

* Update trainer.py

* remove setting property in model

* Update pytorch_lightning/core/lightning.py

Co-authored-by: Rohit Gupta <rohitgr1998@gmail.com>

* Update pytorch_lightning/trainer/trainer.py

Co-authored-by: Rohit Gupta <rohitgr1998@gmail.com>

* Update trainer.py

Co-authored-by: Sean Naren <sean.narenthiran@gmail.com>
Co-authored-by: Roger Shieh <sh.rog@protonmail.ch>
Co-authored-by: Rohit Gupta <rohitgr1998@gmail.com>
Co-authored-by: Jeff Yang <ydcjeff@outlook.com>

pytorch_lightning/core/lightning.py

@@ -164,6 +164,13 @@ class LightningModule(
         """
         return self.device.type == "cuda"
 
+    @property
+    def automatic_optimization(self) -> bool:
+        """
+        If False you are responsible for calling .backward, .step, zero_grad.
+        """
+        return True
+
     def print(self, *args, **kwargs) -> None:
         r"""
         Prints only from process 0. Use this in any distributed mode to log only once.

pytorch_lightning/trainer/connectors/model_connector.py

@@ -35,6 +35,9 @@ class ModelConnector:
         else:
             ref_model = model
 
+        automatic_optimization = ref_model.automatic_optimization and self.trainer.train_loop.automatic_optimization
+        self.trainer.train_loop.automatic_optimization = automatic_optimization
+
         for m in [model, ref_model]:
             m.trainer = self.trainer
             m.logger = self.trainer.logger

pytorch_lightning/trainer/trainer.py

@@ -137,7 +137,7 @@ class Trainer(
         amp_backend: str = 'native',
         amp_level: str = 'O2',
         distributed_backend: Optional[str] = None,
-        automatic_optimization: bool = True,
+        automatic_optimization: Optional[bool] = None,
         move_metrics_to_cpu: bool = False,
     ):
         r"""
@@ -212,7 +212,9 @@ class Trainer(
             log_every_n_steps: How often to log within steps (defaults to every 50 steps).
 
             automatic_optimization: If False you are responsible for calling .backward, .step, zero_grad.
-                Meant to be used with multiple optimizers by advanced users.
+                If False you are responsible for calling .backward, .step, zero_grad in LightningModule.
+                This argument has been moved to LightningModule. It is deprecated here in v1.1 and
+                will be removed in v1.3.
 
             prepare_data_per_node: If True, each LOCAL_RANK=0 will call prepare data.
                 Otherwise only NODE_RANK=0, LOCAL_RANK=0 will prepare data
@@ -355,6 +357,14 @@ class Trainer(
         )
 
         # init train loop related flags
+        # TODO: deprecate in 1.2.0
+        if automatic_optimization is None:
+            automatic_optimization = True
+        else:
+            rank_zero_warn(
+                "Disable automatic optimization with the trainer flag is deprecated and will be removed in v1.3.0!"
+                "Please use the property on the LightningModule for disabling automatic optimization"
+            )
         self.train_loop.on_trainer_init(
             max_epochs,
             min_epochs,