Check if optimizer supports closure (#4981)

* check if optimizer support closure

* cleanup test

* resolve tests

* resolve flake

* update test due to patch limit

* update

* update dep

* Update tests/core/test_lightning_optimizer.py

Co-authored-by: Rohit Gupta <[email protected]>

* Update tests/core/test_lightning_optimizer.py

Co-authored-by: Rohit Gupta <[email protected]>

* resolve bug

* update test

* resolve tests

* Update requirements/extra.txt

Co-authored-by: Jirka Borovec <[email protected]>

* remove bolts dep

* remove bolts

* add missing bolts dep for tests

* remove need for bolts

Co-authored-by: Rohit Gupta <[email protected]>
Co-authored-by: Jirka Borovec <[email protected]>

Author: 3 people

pytorch_lightning/core/optimizer.py

@@ -11,6 +11,7 @@
 # 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.
+import inspect
 import types
 from typing import Any, Callable, Optional
 from weakref import proxy
@@ -60,7 +61,7 @@ def __init__(self,
         self._trainer = None
         self._optimizer = optimizer
         self._accumulate_grad_batches = accumulate_grad_batches
-        self._automatic_optimization = None
+        self._support_closure = 'closure' in inspect.signature(optimizer.step).parameters
         self._optimizer_idx = None
 
     @property
@@ -73,7 +74,6 @@ def accumulate_grad_batches(self, accumulate_grad_batches):
 
     def _on_trainer_init(self, trainer):
         self._trainer = proxy(trainer)
-        self._automatic_optimization = trainer.train_loop.automatic_optimization
         for opt_idx, opt in enumerate(trainer.optimizers):
             if opt == self._optimizer:
                 self._optimizer_idx = opt_idx
@@ -111,7 +111,11 @@ def __optimizer_step(self, *args, closure: Optional[Callable] = None, profiler_n
 
         else:
             with trainer.profiler.profile(profiler_name):
-                optimizer.step(closure=closure, *args, **kwargs)
+                if self._support_closure:
+                    optimizer.step(closure=closure, *args, **kwargs)
+                else:
+                    closure()
+                    optimizer.step(*args, **kwargs)
 
         accelerator_backend = trainer.accelerator_backend
         if accelerator_backend is not None and accelerator_backend.rpc_enabled:

pytorch_lightning/utilities/__init__.py

@@ -54,6 +54,7 @@ def _module_available(module_path: str) -> bool:
 OMEGACONF_AVAILABLE = _module_available("omegaconf")
 HYDRA_AVAILABLE = _module_available("hydra")
 HOROVOD_AVAILABLE = _module_available("horovod.torch")
+BOLTS_AVAILABLE = _module_available("pl_bolts")
 
 TPU_AVAILABLE = XLADeviceUtils.tpu_device_exists()
 FAIRSCALE_AVAILABLE = platform.system() != 'Windows' and _module_available('fairscale.nn.data_parallel')

requirements/extra.txt

@@ -7,4 +7,4 @@ torchtext>=0.3.1, <0.7  # TODO: temporary fix fix for compatibility
 onnx>=1.7.0
 onnxruntime>=1.3.0
 hydra-core>=1.0
-https://github.com/PyTorchLightning/fairscale/archive/pl_1.1.0.zip
+https://github.com/PyTorchLightning/fairscale/archive/pl_1.1.0.zip

tests/core/test_lightning_module.py

@@ -55,6 +55,11 @@ def test_automatic_optimization_num_calls(enable_pl_optimizer, tmpdir):
 
         class TestModel(BoringModel):
 
+            def training_step(self, batch, batch_idx, optimizer_idx):
+                output = self.layer(batch)
+                loss = self.loss(batch, output)
+                return {"loss": loss}
+
             def configure_optimizers(self):
                 optimizer = SGD(self.layer.parameters(), lr=0.1)
                 optimizer_2 = Adam(self.layer.parameters(), lr=0.1)

tests/core/test_lightning_optimizer.py

@@ -19,10 +19,12 @@
 import torch.nn as nn
 from torch.optim import Adam, Optimizer
 
+import pytorch_lightning as pl
 from pytorch_lightning import LightningModule, Trainer
+from pytorch_lightning.callbacks import ModelCheckpoint
 from pytorch_lightning.core.optimizer import LightningOptimizer
 from pytorch_lightning.utilities.exceptions import MisconfigurationException
-from tests.base.boring_model import BoringModel, RandomDictDataset, RandomDictStringDataset
+from tests.base.boring_model import BoringModel, RandomDataset, RandomDictDataset, RandomDictStringDataset
 
 
 def test_lightning_optimizer(tmpdir):
@@ -80,8 +82,8 @@ def configure_optimizers(self):
     assert trainer.optimizers[0].__repr__() == expected
 
 
-@patch("torch.optim.Adam.step")
-@patch("torch.optim.SGD.step")
+@patch("torch.optim.Adam.step", autospec=True)
+@patch("torch.optim.SGD.step", autospec=True)
 def test_lightning_optimizer_manual_optimization(mock_sgd_step, mock_adam_step, tmpdir):
     """
     Test that the user can use our LightningOptimizer. Not recommended for now.
@@ -96,13 +98,13 @@ def training_step(self, batch, batch_idx, optimizer_idx=None):
             output = self.layer(batch)
             loss_1 = self.loss(batch, output)
             self.manual_backward(loss_1, opt_1)
-            opt_1.step(idx="1")
+            opt_1.step()
 
             def closure():
                 output = self.layer(batch)
                 loss_2 = self.loss(batch, output)
                 self.manual_backward(loss_2, opt_2)
-            opt_2.step(closure=closure, idx="2")
+            opt_2.step(closure=closure)
 
         def configure_optimizers(self):
             optimizer_1 = torch.optim.SGD(self.layer.parameters(), lr=0.1)
@@ -133,8 +135,8 @@ def automatic_optimization(self) -> bool:
     assert len(mock_adam_step.mock_calls) == 8
 
 
-@patch("torch.optim.Adam.step")
-@patch("torch.optim.SGD.step")
+@patch("torch.optim.Adam.step", autospec=True)
+@patch("torch.optim.SGD.step", autospec=True)
 def test_lightning_optimizer_manual_optimization_and_accumulated_gradients(mock_sgd_step, mock_adam_step, tmpdir):
     """
     Test that the user can use our LightningOptimizer. Not recommended.
@@ -149,13 +151,13 @@ def training_step(self, batch, batch_idx, optimizer_idx=None):
             output = self.layer(batch)
             loss_1 = self.loss(batch, output)
             self.manual_backward(loss_1, opt_1)
-            opt_1.step(idx="1")
+            opt_1.step()
 
             def closure():
                 output = self.layer(batch)
                 loss_2 = self.loss(batch, output)
                 self.manual_backward(loss_2, opt_2)
-            opt_2.step(closure=closure, idx="2")
+            opt_2.step(closure=closure)
 
         def configure_optimizers(self):
             optimizer_1 = torch.optim.SGD(self.layer.parameters(), lr=0.1)
@@ -195,9 +197,8 @@ def test_state(tmpdir):
     assert isinstance(lightning_optimizer, Adam)
     assert isinstance(lightning_optimizer, Optimizer)
     lightning_dict = {}
-    special_attrs = ["_accumulate_grad_batches", "_optimizer", "_optimizer_idx",
-                     "_trainer", "_use_accumulate_grad_batches_from_trainer", "_automatic_optimization",
-                     "_accumulate_grad_batches"]
+    special_attrs = ["_accumulate_grad_batches", "_optimizer", "_optimizer_idx", "_support_closure",
+                     "_trainer"]
     for k, v in lightning_optimizer.__dict__.items():
         if k not in special_attrs:
             lightning_dict[k] = v
@@ -206,6 +207,55 @@ def test_state(tmpdir):
     assert optimizer.state == lightning_optimizer.state
 
 
+def test_lightning_optimizer_with_wrong_optimizer_interface(tmpdir):
+    class OptimizerWrapper(object):
+        def __init__(self, optimizer):
+            self.optim = optimizer
+            self.state_dict = self.optim.state_dict
+            self.load_state_dict = self.optim.load_state_dict
+            self.zero_grad = self.optim.zero_grad
+            self.add_param_group = self.optim.add_param_group
+            self.__setstate__ = self.optim.__setstate__
+            self.__getstate__ = self.optim.__getstate__
+            self.__repr__ = self.optim.__repr__
+
+        @property
+        def __class__(self):
+            return Optimizer
+
+        @property
+        def state(self):
+            return self.optim.state
+
+        @property
+        def param_groups(self):
+            return self.optim.param_groups
+
+        @param_groups.setter
+        def param_groups(self, value):
+            self.optim.param_groups = value
+
+        def step(self):
+            # wrongly defined step. Should contain closure
+            self.optim.step(closure=None)
+
+    class TestLightningOptimizerModel(BoringModel):
+
+        def configure_optimizers(self):
+            optimizer = torch.optim.Adam(self.parameters(), lr=0.1)
+            optimizer = OptimizerWrapper(optimizer)
+            return [optimizer]
+
+    model = TestLightningOptimizerModel()
+    trainer = Trainer(
+        default_root_dir=tmpdir,
+        max_epochs=1,
+        weights_summary=None,
+        log_every_n_steps=1,
+    )
+    trainer.fit(model)
+
+
 def test_lightning_optimizer_automatic_optimization(tmpdir):
     """
     Test lightning optimize works with make_optimizer_step in automatic_optimization

tests/trainer/optimization/test_manual_optimization.py

@@ -825,7 +825,7 @@ def optimizer_closure():
                     retain_graph = num_backward != backward_idx # noqa E225
                     self.manual_backward(loss_1, opt, retain_graph=retain_graph)
 
-            opt.step(1, closure=optimizer_closure, something="new")
+            opt.step(closure=optimizer_closure)
 
         def training_epoch_end(self, outputs) -> None:
             # outputs should be an array with an entry per optimizer
@@ -855,7 +855,7 @@ def automatic_optimization(self) -> bool:
     )
 
     trainer.fit(model)
-    expected_calls = [call(1, closure=ANY, something="new") for s in range(2)]
+    expected_calls = [call() for s in range(2)]
     step_mock.assert_has_calls(expected_calls)
 
 
@@ -902,7 +902,7 @@ def dis_closure():
             if batch_idx % 4 == 0 :
                 # Note: Set make_optimizer_step to True or it will use by default
                 # Trainer(accumulate_grad_batches=x)
-                opt_dis.step(closure=dis_closure, make_optimizer_step=True, optim='adam')
+                opt_dis.step(closure=dis_closure, make_optimizer_step=True)
 
         def training_epoch_end(self, outputs) -> None:
             # outputs should be an array with an entry per optimizer
@@ -933,10 +933,9 @@ def automatic_optimization(self) -> bool:
     )
 
     trainer.fit(model)
-    expected_calls = [call(closure=ANY, optim='sgd') for s in range(4)]
+    expected_calls = [call(optim='sgd') for s in range(4)]
     mock_sgd_step.assert_has_calls(expected_calls)
-
-    expected_calls = [call(closure=ANY, optim='adam') for s in range(2)]
+    expected_calls = [call() for s in range(2)]
     mock_adam_step.assert_has_calls(expected_calls)