Merge branch 'jdb78:master' into weight-size-bug-fix

.env

@@ -0,0 +1 @@
+PYTORCH_ENABLE_MPS_FALLBACK=1

.github/workflows/test.yml

@@ -72,9 +72,9 @@ jobs:
         shell: bash
         run: poetry install -E "github-actions graph mqf2"
 
-      - name: Install pytorch geometric dependencies
-        shell: bash
-        run: poetry run pip install torch-scatter torch-sparse torch-cluster torch-spline-conv torch-geometric -f https://data.pyg.org/whl/torch-1.12.1+cpu.html
+      # - name: Install pytorch geometric dependencies
+      #   shell: bash
+      #   run: poetry run pip install pyg_lib torch-scatter torch-sparse torch-cluster torch-spline-conv torch-geometric -f https://data.pyg.org/whl/torch-2.0.0+cpu.html
 
       - name: Run pytest
         shell: bash

.gitignore

@@ -104,7 +104,7 @@ celerybeat.pid
 *.sage.py
 
 # Environments
-.env
+# .env
 .venv
 env/
 venv/

.pre-commit-config.yaml

@@ -2,26 +2,26 @@
 # See https://pre-commit.com/hooks.html for more hooks
 repos:
   - repo: https://github.com/pre-commit/pre-commit-hooks
-    rev: v4.1.0
+    rev: v4.4.0
     hooks:
-    - id: trailing-whitespace
-    - id: end-of-file-fixer
-    - id: check-yaml
-    - id: check-ast
-  - repo: https://gitlab.com/pycqa/flake8
-    rev: "3.9.2"
+      - id: trailing-whitespace
+      - id: end-of-file-fixer
+      - id: check-yaml
+      - id: check-ast
+  - repo: https://github.com/pycqa/flake8
+    rev: 6.0.0
     hooks:
-    - id: flake8
+      - id: flake8
   - repo: https://github.com/pre-commit/mirrors-isort
     rev: v5.10.1
     hooks:
-    - id: isort
+      - id: isort
   - repo: https://github.com/psf/black
-    rev: 22.3.0
+    rev: 23.1.0
     hooks:
-    - id: black
+      - id: black
   - repo: https://github.com/nbQA-dev/nbQA
-    rev: 1.3.1
+    rev: 1.6.4
     hooks:
       - id: nbqa-black
       - id: nbqa-isort

CHANGELOG.md

@@ -1,6 +1,16 @@
 # Release Notes
 
-## v0.10.4 UNRELEASED (xx/xx/xxxx)
+## v1.0.0 Update to pytorch 2.0 (10/04/2023)
+
+
+### Breaking Changes
+
+- Upgraded to pytorch 2.0 and lightning 2.0. This brings a couple of changes, such as configuration of trainers. See the [lightning upgrade guide](https://lightning.ai/docs/pytorch/latest/upgrade/migration_guide.html). For PyTorch Forecasting, this particularly means if you are developing own models, the class method `epoch_end` has been renamed to `on_epoch_end` and replacing `model.summarize()` with `ModelSummary(model, max_depth=-1)` and `Tuner(trainer)` is its own class, so `trainer.tuner` needs replacing. (#1280)
+- Changed the `predict()` interface returning named tuple - see tutorials.
+
+### Changes
+
+- The predict method is now using the lightning predict functionality and allows writing results to disk (#1280).
 
 ### Fixed
 
@@ -81,7 +91,7 @@
 
 ### Added
 
-- Added support for running `pytorch_lightning.trainer.test` (#759)
+- Added support for running `lightning.trainer.test` (#759)
 
 ### Fixed
 
@@ -402,7 +412,7 @@ This release has only one purpose: Allow usage of PyTorch Lightning 1.0 - all te
   - Using `LearningRateMonitor` instead of `LearningRateLogger`
   - Use `EarlyStopping` callback in trainer `callbacks` instead of `early_stopping` argument
   - Update metric system `update()` and `compute()` methods
-  - Use `trainer.tuner.lr_find()` instead of `trainer.lr_find()` in tutorials and examples
+  - Use `Tuner(trainer).lr_find()` instead of `trainer.lr_find()` in tutorials and examples
 - Update poetry to 1.1.0
 
 ---

README.md

@@ -34,7 +34,6 @@ Specifically, the package provides
 - Multiple neural network architectures for timeseries forecasting that have been enhanced
   for real-world deployment and come with in-built interpretation capabilities
 - Multi-horizon timeseries metrics
-- Ranger optimizer for faster model training
 - Hyperparameter tuning with [optuna](https://optuna.readthedocs.io/)
 
 The package is built on [pytorch-lightning](https://pytorch-lightning.readthedocs.io/) to allow training on CPUs, single and multiple GPUs out-of-the-box.
@@ -86,11 +85,12 @@ Networks can be trained with the [PyTorch Lighning Trainer](https://pytorch-ligh
 
 ```python
 # imports for training
-import pytorch_lightning as pl
-from pytorch_lightning.loggers import TensorBoardLogger
-from pytorch_lightning.callbacks import EarlyStopping, LearningRateMonitor
+import lightning.pytorch as pl
+from lightning.pytorch.loggers import TensorBoardLogger
+from lightning.pytorch.callbacks import EarlyStopping, LearningRateMonitor
 # import dataset, network to train and metric to optimize
 from pytorch_forecasting import TimeSeriesDataSet, TemporalFusionTransformer, QuantileLoss
+from lightning.pytorch.tuner import Tuner
 
 # load data: this is pandas dataframe with at least a column for
 # * the target (what you want to predict)
@@ -133,7 +133,7 @@ early_stop_callback = EarlyStopping(monitor="val_loss", min_delta=1e-4, patience
 lr_logger = LearningRateMonitor()
 trainer = pl.Trainer(
     max_epochs=100,
-    gpus=0,  # run on CPU, if on multiple GPUs, use accelerator="ddp"
+    accelerator="auto",  # run on CPU, if on multiple GPUs, use strategy="ddp"
     gradient_clip_val=0.1,
     limit_train_batches=30,  # 30 batches per epoch
     callbacks=[lr_logger, early_stop_callback],
@@ -160,7 +160,7 @@ tft = TemporalFusionTransformer.from_dataset(
 print(f"Number of parameters in network: {tft.size()/1e3:.1f}k")
 
 # find the optimal learning rate
-res = trainer.lr_find(
+res = Tuner(trainer).lr_find(
     tft, train_dataloaders=train_dataloader, val_dataloaders=val_dataloader, early_stop_threshold=1000.0, max_lr=0.3,
 )
 # and plot the result - always visually confirm that the suggested learning rate makes sense

docs/requirements.txt

@@ -3,15 +3,17 @@ nbsphinx
 pandoc
 docutils
 pydata-sphinx-theme
-pytorch-lightning>=0.9.0
+lightning>=2.0.0
 cloudpickle
-torch>=1.6
-optuna>=2.0
+torch>=2.0
+optuna>=3.1.0
 scipy
-pandas>=1.0
-scikit-learn>0.23
+pandas>=1.3
+scikit-learn>1.2
 matplotlib
 statsmodels
 ipython
 nbconvert>=6.3.0
 recommonmark>=0.7.1
+pytorch-optimizer>=2.5.1
+fastapi>0.80

docs/source/api.rst

@@ -11,5 +11,4 @@ API
    data
    models
    metrics
-   optim
    utils

docs/source/getting-started.rst

@@ -46,7 +46,7 @@ The general setup for training and testing a model is
    directly if you do not wish to load the entire training dataset at inference time.
 
 #. Instantiate a model using the its ``.from_dataset()`` method.
-#. Create a ``pytorch_lightning.Trainer()`` object.
+#. 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
    to understand if it has converged with acceptable accuracy.
@@ -65,9 +65,9 @@ Example
 
 .. code-block:: python
 
-    import pytorch_lightning as pl
-    from pytorch_lightning.callbacks import EarlyStopping, LearningRateMonitor
-
+    import lightning.pytorch as pl
+    from lightning.pytorch.callbacks import EarlyStopping, LearningRateMonitor
+    from lightning.pytorch.tuner import Tuner
     from pytorch_forecasting import TimeSeriesDataSet, TemporalFusionTransformer
 
     # load data
@@ -105,7 +105,7 @@ Example
     lr_logger = LearningRateMonitor()
     trainer = pl.Trainer(
         max_epochs=100,
-        gpus=0,
+        accelerator="auto",
         gradient_clip_val=0.1,
         limit_train_batches=30,
         callbacks=[lr_logger, early_stop_callback],
@@ -127,7 +127,7 @@ Example
     print(f"Number of parameters in network: {tft.size()/1e3:.1f}k")
 
     # find optimal learning rate (set limit_train_batches to 1.0 and log_interval = -1)
-    res = trainer.tuner.lr_find(
+    res = Tuner(trainer).lr_find(
         tft, train_dataloaders=train_dataloader, val_dataloaders=val_dataloader, early_stop_threshold=1000.0, max_lr=0.3,
     )
 

docs/source/index.rst

@@ -27,7 +27,6 @@ Specifically, the package provides
 * Multiple neural network architectures for timeseries forecasting that have been enhanced
   for real-world deployment and come with in-built interpretation capabilities
 * Multi-horizon timeseries metrics
-* Ranger optimizer for faster model training
 * Hyperparameter tuning with `optuna <https://optuna.readthedocs.io/>`_
 
 The package is built on `PyTorch Lightning <https://pytorch-lightning.readthedocs.io/>`_ to allow