Fixed typo "monotone_constaints"

pytorch_forecasting/models/base_model.py

@@ -483,7 +483,7 @@ def __init__(
         reduce_on_plateau_min_lr: float = 1e-5,
         weight_decay: float = 0.0,
         optimizer_params: Dict[str, Any] = None,
-        monotone_constaints: Dict[str, int] = None,
+        monotone_constraints: Dict[str, int] = {},
         output_transformer: Callable = None,
         optimizer=None,
     ):
@@ -508,7 +508,7 @@ def __init__(
                 Defaults to 1e-5
             weight_decay (float): weight decay. Defaults to 0.0.
             optimizer_params (Dict[str, Any]): additional parameters for the optimizer. Defaults to {}.
-            monotone_constaints (Dict[str, int]): dictionary of monotonicity constraints for continuous decoder
+            monotone_constraints (Dict[str, int]): dictionary of monotonicity constraints for continuous decoder
                 variables mapping
                 position (e.g. ``"0"`` for first position) to constraint (``-1`` for negative and ``+1`` for positive,
                 larger numbers add more weight to the constraint vs. the loss but are usually not necessary).
@@ -522,8 +522,8 @@ def __init__(
                 `"ranger" <https://pytorch-optimizers.readthedocs.io/en/latest/optimizer_api.html#ranger21>`_,
                 if pytorch_optimizer is installed, otherwise "adam".
         """  # noqa: E501
-        if monotone_constaints is None:
-            monotone_constaints = {}
+        if monotone_constraints is None:
+            monotone_constraints = {}
         super().__init__()
         # update hparams
         frame = inspect.currentframe()
@@ -883,7 +883,7 @@ def step(
                     y[1],
                 )
 
-        if self.training and len(self.hparams.monotone_constaints) > 0:
+        if self.training and len(self.hparams.monotone_constraints) > 0:
             # calculate gradient with respect to continous decoder features
             x["decoder_cont"].requires_grad_(True)
             assert not torch._C._get_cudnn_enabled(), (
@@ -913,11 +913,11 @@ def step(
             indices = torch.tensor(
                 [
                     self.hparams.x_reals.index(name)
-                    for name in self.hparams.monotone_constaints.keys()
+                    for name in self.hparams.monotone_constraints.keys()
                 ]
             )
             monotonicity = torch.tensor(
-                list(self.hparams.monotone_constaints.values()),
+                list(self.hparams.monotone_constraints.values()),
                 dtype=gradient.dtype,
                 device=gradient.device,
             )

tests/test_models/test_temporal_fusion_transformer.py

@@ -139,10 +139,10 @@ def _integration(dataloader, tmp_path, loss=None, trainer_kwargs=None, **kwargs)
     )
     # test monotone constraints automatically
     if "discount_in_percent" in train_dataloader.dataset.reals:
-        monotone_constaints = {"discount_in_percent": +1}
+        monotone_constraints = {"discount_in_percent": +1}
         cuda_context = torch.backends.cudnn.flags(enabled=False)
     else:
-        monotone_constaints = {}
+        monotone_constraints = {}
         cuda_context = nullcontext()
 
     kwargs.setdefault("learning_rate", 0.15)
@@ -175,7 +175,7 @@ def _integration(dataloader, tmp_path, loss=None, trainer_kwargs=None, **kwargs)
             log_interval=5,
             log_val_interval=1,
             log_gradient_flow=True,
-            monotone_constaints=monotone_constaints,
+            monotone_constraints=monotone_constraints,
             **kwargs,
         )
         net.size()