fix missing call to untoggle_optimizer when accumulating gradients (#8284)

* add fix

* toggle test

* re-structure

* update changelog

* update comment

* remove debugging assertion

CHANGELOG.md

@@ -358,6 +358,9 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Fixed passing a custom `DDPPlugin` when choosing `accelerator="ddp_cpu"` for the accelerator ([#6208](https://github.com/PyTorchLightning/pytorch-lightning/pull/6208))
 
 
+- Fixed missing call to `LightningModule.untoggle_optimizer` in training loop when running gradient accumulation with multiple optimizers ([#8284](https://github.com/PyTorchLightning/pytorch-lightning/pull/8284))
+
+
 ## [1.3.8] - 2021-07-01
 
 ### Fixed

pytorch_lightning/loops/batch/training_batch_loop.py

@@ -204,20 +204,17 @@ class TrainingBatchLoop(Loop):
         else:
             if self.trainer.lightning_module.automatic_optimization:
                 self._optimizer_step(optimizer, opt_idx, batch_idx, closure)
-                if len(self.trainer.optimizers) > 1:
-                    # revert back to previous state
-                    self.trainer.lightning_module.untoggle_optimizer(opt_idx)
             else:
                 result = self._training_step(split_batch, batch_idx, opt_idx, self._hiddens)
 
-            if not result:
-                # user decided to skip optimization
-                return result
-
-            # update running loss + reset accumulated loss
+        if result:
+            # if no result, user decided to skip optimization
+            # otherwise update running loss + reset accumulated loss
             self._update_running_loss(result.loss)
+            self._process_closure_result(result)
 
-        self._process_closure_result(result)
+        # untoggle model params
+        self._run_optimization_end(opt_idx)
         return result
 
     def _training_step_and_backward_closure(
@@ -509,6 +506,11 @@ class TrainingBatchLoop(Loop):
             model = self.trainer.lightning_module
             model.toggle_optimizer(optimizer, opt_idx)
 
+    def _run_optimization_end(self, opt_idx: int) -> None:
+        if self.trainer.lightning_module.automatic_optimization and len(self.trainer.optimizers) > 1:
+            model = self.trainer.lightning_module
+            model.untoggle_optimizer(opt_idx)
+
     @contextmanager
     def block_ddp_sync_behaviour(self, should_block_sync: bool = False) -> Generator[None, None, None]:
         """

tests/core/test_lightning_module.py

@@ -197,7 +197,7 @@ def test_toggle_untoggle_2_optimizers_no_shared_parameters(tmpdir):
         max_epochs=1,
         default_root_dir=tmpdir,
         limit_train_batches=8,
-        accumulate_grad_batches=1,
+        accumulate_grad_batches=2,
         limit_val_batches=0,
     )
     trainer.fit(model)
@@ -331,7 +331,7 @@ def test_toggle_untoggle_3_optimizers_shared_parameters(tmpdir):
         max_epochs=1,
         default_root_dir=tmpdir,
         limit_train_batches=8,
-        accumulate_grad_batches=1,
+        accumulate_grad_batches=2,
     )
 
     trainer.fit(model)