Merge pull request #97 from jdb78/maintenance/pytorch-lightning-1.0.0rc4

Rename LearningRateLogger and reposition EarlyStopping callback for l…

Author: jdb78

README.md

@@ -54,7 +54,7 @@ documentation with detailed tutorials.
 
 ```python
 import pytorch_lightning as pl
-from pytorch_lightning.callbacks import EarlyStopping
+from pytorch_lightning.callbacks import EarlyStopping, LearningRateMonitor
 
 from pytorch_forecasting import TimeSeriesDataSet, TemporalFusionTransformer
 
@@ -89,14 +89,13 @@ val_dataloader = validation.to_dataloader(train=False, batch_size=batch_size, nu
 
 
 early_stop_callback = EarlyStopping(monitor="val_loss", min_delta=1e-4, patience=1, verbose=False, mode="min")
-lr_logger = LearningRateLogger()
+lr_logger = LearningRateMonitor()
 trainer = pl.Trainer(
     max_epochs=100,
     gpus=0,
     gradient_clip_val=0.1,
-    early_stop_callback=early_stop_callback,
     limit_train_batches=30,
-    callbacks=[lr_logger],
+    callbacks=[lr_logger, early_stop_callback],
 )
 
 

docs/source/getting-started.rst

@@ -62,7 +62,7 @@ Example
 .. code-block:: python
 
     import pytorch_lightning as pl
-    from pytorch_lightning.callbacks import EarlyStopping
+    from pytorch_lightning.callbacks import EarlyStopping, LearningRateMonitor
 
     from pytorch_forecasting import TimeSeriesDataSet, TemporalFusionTransformer
 
@@ -98,14 +98,13 @@ Example
 
     # define trainer with early stopping
     early_stop_callback = EarlyStopping(monitor="val_loss", min_delta=1e-4, patience=1, verbose=False, mode="min")
-    lr_logger = LearningRateLogger()
+    lr_logger = LearningRateMonitor()
     trainer = pl.Trainer(
         max_epochs=100,
         gpus=0,
         gradient_clip_val=0.1,
-        early_stop_callback=early_stop_callback,
         limit_train_batches=30,
-        callbacks=[lr_logger],
+        callbacks=[lr_logger, early_stop_callback],
     )
 
     # create the model

docs/source/tutorials/ar.ipynb

@@ -723,7 +723,7 @@
     "    gpus=0,\n",
     "    weights_summary=\"top\",\n",
     "    gradient_clip_val=0.1,\n",
-    "    early_stop_callback=early_stop_callback,\n",
+    "    callbacks=[early_stop_callback],\n",
     "    limit_train_batches=30,\n",
     ")\n",
     "\n",
@@ -1097,7 +1097,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.7.7"
+   "version": "3.8.3"
   }
  },
  "nbformat": 4,

docs/source/tutorials/stallion.ipynb

@@ -6,7 +6,7 @@
 import pandas as pd
 from pandas.core.common import SettingWithCopyWarning
 import pytorch_lightning as pl
-from pytorch_lightning.callbacks import EarlyStopping, LearningRateLogger
+from pytorch_lightning.callbacks import EarlyStopping, LearningRateMonitor
 from pytorch_lightning.loggers import TensorBoardLogger
 import torch
 
@@ -64,20 +64,19 @@
 validation.save("validation.pkl")
 
 early_stop_callback = EarlyStopping(monitor="val_loss", min_delta=1e-4, patience=5, verbose=False, mode="min")
-lr_logger = LearningRateLogger()
+lr_logger = LearningRateMonitor()
 
 trainer = pl.Trainer(
     max_epochs=100,
     gpus=0,
     weights_summary="top",
     gradient_clip_val=0.1,
-    early_stop_callback=early_stop_callback,
     limit_train_batches=30,
     limit_val_batches=3,
     # fast_dev_run=True,
     # logger=logger,
     # profiler=True,
-    callbacks=[lr_logger],
+    callbacks=[lr_logger, early_stop_callback],
 )
 
 

examples/ar.py

@@ -55,7 +55,7 @@
     gpus=0,
     weights_summary="top",
     gradient_clip_val=0.1,
-    early_stop_callback=early_stop_callback,
+    callbacks=[early_stop_callback],
     limit_train_batches=15,
     # limit_val_batches=1,
     # fast_dev_run=True,

examples/nbeats.py

@@ -6,7 +6,7 @@
 import pandas as pd
 from pandas.core.common import SettingWithCopyWarning
 import pytorch_lightning as pl
-from pytorch_lightning.callbacks import EarlyStopping, LearningRateLogger
+from pytorch_lightning.callbacks import EarlyStopping, LearningRateMonitor
 from pytorch_lightning.loggers import TensorBoardLogger
 import torch
 
@@ -93,21 +93,20 @@
 validation.save("validation.pkl")
 
 early_stop_callback = EarlyStopping(monitor="val_loss", min_delta=1e-4, patience=10, verbose=False, mode="min")
-lr_logger = LearningRateLogger()
+lr_logger = LearningRateMonitor()
 
 trainer = pl.Trainer(
     max_epochs=100,
     gpus=0,
     weights_summary="top",
     gradient_clip_val=0.1,
-    early_stop_callback=early_stop_callback,
     limit_train_batches=30,
     # val_check_interval=20,
     # limit_val_batches=1,
     # fast_dev_run=True,
     # logger=logger,
     # profiler=True,
-    callbacks=[lr_logger],
+    callbacks=[lr_logger, early_stop_callback],
 )
 
 

examples/stallion.py

@@ -9,7 +9,7 @@
 from optuna.integration import PyTorchLightningPruningCallback, TensorBoardCallback
 import pytorch_lightning as pl
 from pytorch_lightning import Callback
-from pytorch_lightning.callbacks import LearningRateLogger
+from pytorch_lightning.callbacks import LearningRateMonitor
 from pytorch_lightning.loggers import TensorBoardLogger
 import statsmodels.api as sm
 import torch
@@ -99,16 +99,19 @@ def objective(trial: optuna.Trial) -> float:
         # TensorBoard. We don't use any logger here as it requires us to implement several abstract
         # methods. Instead we setup a simple callback, that saves metrics from each validation step.
         metrics_callback = MetricsCallback()
-        learning_rate_callback = LearningRateLogger()
+        learning_rate_callback = LearningRateMonitor()
         logger = TensorBoardLogger(log_dir, name="optuna", version=trial.number)
         gradient_clip_val = trial.suggest_loguniform("gradient_clip_val", *gradient_clip_val_range)
         trainer = pl.Trainer(
             checkpoint_callback=checkpoint_callback,
             max_epochs=max_epochs,
             gradient_clip_val=gradient_clip_val,
             gpus=[0] if torch.cuda.is_available() else None,
-            callbacks=[metrics_callback, learning_rate_callback],
-            early_stop_callback=PyTorchLightningPruningCallback(trial, monitor="val_loss"),
+            callbacks=[
+                metrics_callback,
+                learning_rate_callback,
+                PyTorchLightningPruningCallback(trial, monitor="val_loss"),
+            ],
             logger=logger,
             **trainer_kwargs,
         )

pytorch_forecasting/models/temporal_fusion_transformer/tuning.py

@@ -4,7 +4,6 @@
 from pytorch_lightning.callbacks import EarlyStopping, ModelCheckpoint
 from pytorch_lightning.loggers import TensorBoardLogger
 
-from pytorch_forecasting.metrics import QuantileLoss
 from pytorch_forecasting.models import NBeats
 
 
@@ -21,7 +20,7 @@ def test_integration(dataloaders_fixed_window_without_coveratiates, tmp_path, gp
         gpus=gpus,
         weights_summary="top",
         gradient_clip_val=0.1,
-        early_stop_callback=early_stop_callback,
+        callbacks=[early_stop_callback],
         fast_dev_run=True,
         logger=logger,
     )

tests/test_models/test_nbeats.py

@@ -40,7 +40,7 @@ def test_integration(multiple_dataloaders_with_coveratiates, tmp_path, gpus):
         gpus=gpus,
         weights_summary="top",
         gradient_clip_val=0.1,
-        early_stop_callback=early_stop_callback,
+        callbacks=[early_stop_callback],
         fast_dev_run=True,
         logger=logger,
     )