save initial arguments (#4163)

* save initial arguments

* typing

* chlog

* .

CHANGELOG.md

@@ -21,6 +21,8 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 
 ### Fixed
 
+- Fixed `hparams` saving - save the state when `save_hyperparameters()` is called [in `__init__`] ([#4163](https://github.com/PyTorchLightning/pytorch-lightning/pull/4163))
+
 
 
 ## [1.0.1] - 2020-10-14

pytorch_lightning/core/lightning.py

@@ -13,6 +13,7 @@
 # limitations under the License.
 
 import collections
+import copy
 import inspect
 import os
 import re
@@ -1448,9 +1449,11 @@ def save_hyperparameters(self, *args, frame=None) -> None:
         init_args = get_init_args(frame)
         assert init_args, "failed to inspect the self init"
         if not args:
+            # take all arguments
             hp = init_args
             self._hparams_name = "kwargs" if hp else None
         else:
+            # take only listed arguments in `save_hparams`
             isx_non_str = [i for i, arg in enumerate(args) if not isinstance(arg, str)]
             if len(isx_non_str) == 1:
                 hp = args[isx_non_str[0]]
@@ -1463,6 +1466,8 @@ def save_hyperparameters(self, *args, frame=None) -> None:
         # `hparams` are expected here
         if hp:
             self._set_hparams(hp)
+        # make deep copy so  there is not other runtime changes reflected
+        self._hparams_initial = copy.deepcopy(self._hparams)
 
     def _set_hparams(self, hp: Union[dict, Namespace, str]) -> None:
         if isinstance(hp, Namespace):
@@ -1594,11 +1599,18 @@ def to_torchscript(
         return torchscript_module
 
     @property
-    def hparams(self) -> Union[AttributeDict, str]:
+    def hparams(self) -> Union[AttributeDict, dict, Namespace]:
         if not hasattr(self, "_hparams"):
             self._hparams = AttributeDict()
         return self._hparams
 
+    @property
+    def hparams_initial(self) -> AttributeDict:
+        if not hasattr(self, "_hparams_initial"):
+            self._hparams_initial = AttributeDict()
+        # prevent any change
+        return copy.deepcopy(self._hparams_initial)
+
     @hparams.setter
     def hparams(self, hp: Union[dict, Namespace, Any]):
         hparams_assignment_name = self.__get_hparams_assignment_variable()

pytorch_lightning/trainer/training_loop.py

@@ -129,7 +129,7 @@ def setup_training(self, model: LightningModule):
         # log hyper-parameters
         if self.trainer.logger is not None:
             # save exp to get started (this is where the first experiment logs are written)
-            self.trainer.logger.log_hyperparams(ref_model.hparams)
+            self.trainer.logger.log_hyperparams(ref_model.hparams_initial)
             self.trainer.logger.log_graph(ref_model)
             self.trainer.logger.save()
 

tests/models/test_hparams.py

@@ -25,7 +25,7 @@
 from pytorch_lightning import Trainer, LightningModule
 from pytorch_lightning.core.saving import save_hparams_to_yaml, load_hparams_from_yaml
 from pytorch_lightning.utilities import AttributeDict, is_picklable
-from tests.base import EvalModelTemplate, TrialMNIST
+from tests.base import EvalModelTemplate, TrialMNIST, BoringModel
 
 
 class SaveHparamsModel(EvalModelTemplate):
@@ -554,3 +554,28 @@ def test_args(tmpdir):
     with pytest.raises(TypeError, match="__init__\(\) got an unexpected keyword argument 'test'"):
         SubClassVarArgs.load_from_checkpoint(raw_checkpoint_path)
 
+
+class RuntimeParamChangeModel(BoringModel):
+    def __init__(self, running_arg):
+        super().__init__()
+        self.save_hyperparameters()
+
+
+def test_init_arg_with_runtime_change(tmpdir):
+    model = RuntimeParamChangeModel(123)
+    assert model.hparams.running_arg == 123
+    model.hparams.running_arg = -1
+    assert model.hparams.running_arg == -1
+
+    trainer = Trainer(
+        default_root_dir=tmpdir,
+        limit_train_batches=2,
+        limit_val_batches=2,
+        limit_test_batches=2,
+        max_epochs=1,
+    )
+    trainer.fit(model)
+
+    path_yaml = os.path.join(trainer.logger.log_dir, trainer.logger.NAME_HPARAMS_FILE)
+    hparams = load_hparams_from_yaml(path_yaml)
+    assert hparams.get('running_arg') == 123