Merge branch 'sktime:main' into newDataset

Author: phoeenniixx

docs/source/getting-started.rst

@@ -23,12 +23,11 @@ Otherwise, you can proceed with
 
 
 Alternatively, to install the package via ``conda``:
-
 .. code-block:: bash
 
     conda install pytorch-forecasting pytorch>=1.7 -c pytorch -c conda-forge
 
-PyTorch Forecasting is now installed from the conda-forge channel while PyTorch is install from the pytorch channel.
+PyTorch Forecasting is now installed from the conda-forge channel while PyTorch is installed from the pytorch channel.
 
 To use the MQF2 loss (multivariate quantile loss), also install
 
@@ -54,7 +53,7 @@ The general setup for training and testing a model is
    Similarly, a test dataset or later a dataset for inference can be created. You can store the dataset parameters
    directly if you do not wish to load the entire training dataset at inference time.
 
-#. Instantiate a model using the its ``.from_dataset()`` method.
+#. Instantiate a model using the ``.from_dataset()`` method.
 #. Create a ``lightning.Trainer()`` object.
 #. Find the optimal learning rate with its ``.tuner.lr_find()`` method.
 #. Train the model with early stopping on the training dataset and use the tensorboard logs
@@ -65,7 +64,7 @@ The general setup for training and testing a model is
 #. Load the model from the model checkpoint and apply it to new data.
 
 
-The :ref:`Tutorials <tutorials>` section provides detailled guidance and examples on how to use models and implement new ones.
+The :ref:`Tutorials <tutorials>` section provides detailed guidance and examples on how to use models and implement new ones.
 
 
 Example

docs/source/tutorials/ar.ipynb

@@ -268,7 +268,7 @@
    ],
    "source": [
     "pl.seed_everything(42)\n",
-    "trainer = pl.Trainer(accelerator=\"auto\", gradient_clip_val=0.01)\n",
+    "trainer = pl.Trainer(accelerator=\"auto\", gradient_clip_val=0.1)\n",
     "net = NBeats.from_dataset(training, learning_rate=3e-2, weight_decay=1e-2, widths=[32, 512], backcast_loss_ratio=0.1)"
    ]
   },
@@ -448,7 +448,7 @@
     "    max_epochs=3,\n",
     "    accelerator=\"auto\",\n",
     "    enable_model_summary=True,\n",
-    "    gradient_clip_val=0.01,\n",
+    "    gradient_clip_val=0.1,\n",
     "    callbacks=[early_stop_callback],\n",
     "    limit_train_batches=150,\n",
     ")\n",

pyproject.toml

@@ -106,7 +106,7 @@ dev = [
 
 # docs - dependencies for building the documentation
 docs = [
-  "sphinx>3.2",
+  "sphinx>3.2,<7.2.6",
   "pydata-sphinx-theme",
   "nbsphinx",
   "pandoc",

pytorch_forecasting/metrics/_mqf2_utils.py

@@ -2,7 +2,6 @@
 
 from typing import List, Optional, Tuple
 
-from cpflows.flows import DeepConvexFlow, SequentialFlow
 import torch
 from torch.distributions import (
     AffineTransform,
@@ -12,6 +11,11 @@
 )
 import torch.nn.functional as F
 
+from pytorch_forecasting.utils._dependencies import _safe_import
+
+DeepConvexFlow = _safe_import("cpflows.flows.DeepConvexFlow")
+SequentialFlow = _safe_import("cpflows.flows.SequentialFlow")
+
 
 class DeepConvexNet(DeepConvexFlow):
     r"""

pytorch_forecasting/models/__init__.py

@@ -2,7 +2,7 @@
 Models for timeseries forecasting.
 """
 
-from pytorch_forecasting.models.base_model import (
+from pytorch_forecasting.models.base import (
     AutoRegressiveBaseModel,
     AutoRegressiveBaseModelWithCovariates,
     BaseModel,

pytorch_forecasting/models/base/__init__.py

@@ -0,0 +1,17 @@
+"""Base classes for pytorch-foercasting models."""
+
+from pytorch_forecasting.models.base._base_model import (
+    AutoRegressiveBaseModel,
+    AutoRegressiveBaseModelWithCovariates,
+    BaseModel,
+    BaseModelWithCovariates,
+    Prediction,
+)
+
+__all__ = [
+    "AutoRegressiveBaseModel",
+    "AutoRegressiveBaseModelWithCovariates",
+    "BaseModel",
+    "BaseModelWithCovariates",
+    "Prediction",
+]