remove TQDMProgressBar callback (#899)

nbs/common.base_auto.ipynb

@@ -68,7 +68,6 @@
     "import torch\n",
     "import pytorch_lightning as pl\n",
     "\n",
-    "from pytorch_lightning.callbacks import TQDMProgressBar\n",
     "from ray import air, tune\n",
     "from ray.tune.integration.pytorch_lightning import TuneReportCallback\n",
     "from ray.tune.search.basic_variant import BasicVariantGenerator"
@@ -254,9 +253,9 @@
     "        `test_size`: int, test size for temporal cross-validation.<br>\n",
     "        \"\"\"\n",
     "        metrics = {\"loss\": \"ptl/val_loss\", \"train_loss\": \"train_loss\"}\n",
-    "        callbacks = [TQDMProgressBar(), TuneReportCallback(metrics, on=\"validation_end\")]\n",
+    "        callbacks = [TuneReportCallback(metrics, on=\"validation_end\")]\n",
     "        if 'callbacks' in config_step.keys():\n",
-    "            callbacks += config_step['callbacks']\n",
+    "            callbacks.extend(config_step['callbacks'])\n",
     "        config_step = {**config_step, **{'callbacks': callbacks}}\n",
     "\n",
     "        # Protect dtypes from tune samplers\n",

nbs/common.base_multivariate.ipynb

@@ -61,7 +61,6 @@
     "import torch\n",
     "import torch.nn as nn\n",
     "import pytorch_lightning as pl\n",
-    "from pytorch_lightning.callbacks import TQDMProgressBar\n",
     "from pytorch_lightning.callbacks.early_stopping import EarlyStopping\n",
     "\n",
     "from neuralforecast.common._scalers import TemporalNorm\n",
@@ -166,14 +165,12 @@
     "            raise Exception('max_epochs is deprecated, use max_steps instead.')\n",
     "\n",
     "        # Callbacks\n",
-    "        if trainer_kwargs.get('callbacks', None) is None:\n",
-    "            callbacks = [TQDMProgressBar()]\n",
-    "            # Early stopping\n",
-    "            if self.early_stop_patience_steps > 0:\n",
-    "                callbacks += [EarlyStopping(monitor='ptl/val_loss',\n",
-    "                                            patience=self.early_stop_patience_steps)]\n",
-    "\n",
-    "            trainer_kwargs['callbacks'] = callbacks\n",
+    "        if 'callbacks' not in trainer_kwargs and self.early_stop_patience_steps > 0:\n",
+    "            trainer_kwargs['callbacks'] = [\n",
+    "                EarlyStopping(\n",
+    "                    monitor='ptl/val_loss', patience=self.early_stop_patience_steps\n",
+    "                )\n",
+    "            ]\n",
     "\n",
     "        # Add GPU accelerator if available\n",
     "        if trainer_kwargs.get('accelerator', None) is None:\n",

nbs/common.base_recurrent.ipynb

@@ -67,7 +67,6 @@
     "import torch\n",
     "import torch.nn as nn\n",
     "import pytorch_lightning as pl\n",
-    "from pytorch_lightning.callbacks import TQDMProgressBar\n",
     "from pytorch_lightning.callbacks.early_stopping import EarlyStopping\n",
     "\n",
     "from neuralforecast.common._scalers import TemporalNorm\n",
@@ -178,14 +177,12 @@
     "            raise Exception('max_epochs is deprecated, use max_steps instead.')\n",
     "\n",
     "        # Callbacks\n",
-    "        if trainer_kwargs.get('callbacks', None) is None:\n",
-    "            callbacks = [TQDMProgressBar()]\n",
-    "            # Early stopping\n",
-    "            if self.early_stop_patience_steps > 0:\n",
-    "                callbacks += [EarlyStopping(monitor='ptl/val_loss',\n",
-    "                                            patience=self.early_stop_patience_steps)]\n",
-    "\n",
-    "            trainer_kwargs['callbacks'] = callbacks\n",
+    "        if 'callbacks' not in trainer_kwargs and self.early_stop_patience_steps > 0:\n",
+    "            trainer_kwargs['callbacks'] = [\n",
+    "                EarlyStopping(\n",
+    "                    monitor='ptl/val_loss', patience=self.early_stop_patience_steps\n",
+    "                )\n",
+    "            ]\n",
     "\n",
     "        # Add GPU accelerator if available\n",
     "        if trainer_kwargs.get('accelerator', None) is None:\n",

nbs/common.base_windows.ipynb

@@ -67,7 +67,6 @@
     "import torch\n",
     "import torch.nn as nn\n",
     "import pytorch_lightning as pl\n",
-    "from pytorch_lightning.callbacks import TQDMProgressBar\n",
     "from pytorch_lightning.callbacks.early_stopping import EarlyStopping\n",
     "\n",
     "from neuralforecast.common._scalers import TemporalNorm\n",
@@ -192,14 +191,12 @@
     "            raise Exception('max_epochs is deprecated, use max_steps instead.')\n",
     "\n",
     "        # Callbacks\n",
-    "        if trainer_kwargs.get('callbacks', None) is None:\n",
-    "            callbacks = [TQDMProgressBar()]\n",
-    "            # Early stopping\n",
-    "            if self.early_stop_patience_steps > 0:\n",
-    "                callbacks += [EarlyStopping(monitor='ptl/val_loss',\n",
-    "                                            patience=self.early_stop_patience_steps)]\n",
-    "\n",
-    "            trainer_kwargs['callbacks'] = callbacks\n",
+    "        if 'callbacks' not in trainer_kwargs and self.early_stop_patience_steps > 0:\n",
+    "            trainer_kwargs['callbacks'] = [\n",
+    "                EarlyStopping(\n",
+    "                    monitor='ptl/val_loss', patience=self.early_stop_patience_steps\n",
+    "                )\n",
+    "            ]\n",
     "\n",
     "        # Add GPU accelerator if available\n",
     "        if trainer_kwargs.get('accelerator', None) is None:\n",

neuralforecast/common/_base_auto.py

@@ -10,7 +10,6 @@
 import torch
 import pytorch_lightning as pl
 
-from pytorch_lightning.callbacks import TQDMProgressBar
 from ray import air, tune
 from ray.tune.integration.pytorch_lightning import TuneReportCallback
 from ray.tune.search.basic_variant import BasicVariantGenerator
@@ -194,12 +193,9 @@ def _train_tune(self, config_step, cls_model, dataset, val_size, test_size):
         `test_size`: int, test size for temporal cross-validation.<br>
         """
         metrics = {"loss": "ptl/val_loss", "train_loss": "train_loss"}
-        callbacks = [
-            TQDMProgressBar(),
-            TuneReportCallback(metrics, on="validation_end"),
-        ]
+        callbacks = [TuneReportCallback(metrics, on="validation_end")]
         if "callbacks" in config_step.keys():
-            callbacks += config_step["callbacks"]
+            callbacks.extend(config_step["callbacks"])
         config_step = {**config_step, **{"callbacks": callbacks}}
 
         # Protect dtypes from tune samplers

neuralforecast/common/_base_multivariate.py

@@ -11,7 +11,6 @@
 import torch
 import torch.nn as nn
 import pytorch_lightning as pl
-from pytorch_lightning.callbacks import TQDMProgressBar
 from pytorch_lightning.callbacks.early_stopping import EarlyStopping
 
 from ._scalers import TemporalNorm
@@ -115,17 +114,12 @@ def __init__(
             raise Exception("max_epochs is deprecated, use max_steps instead.")
 
         # Callbacks
-        if trainer_kwargs.get("callbacks", None) is None:
-            callbacks = [TQDMProgressBar()]
-            # Early stopping
-            if self.early_stop_patience_steps > 0:
-                callbacks += [
-                    EarlyStopping(
-                        monitor="ptl/val_loss", patience=self.early_stop_patience_steps
-                    )
-                ]
-
-            trainer_kwargs["callbacks"] = callbacks
+        if "callbacks" not in trainer_kwargs and self.early_stop_patience_steps > 0:
+            trainer_kwargs["callbacks"] = [
+                EarlyStopping(
+                    monitor="ptl/val_loss", patience=self.early_stop_patience_steps
+                )
+            ]
 
         # Add GPU accelerator if available
         if trainer_kwargs.get("accelerator", None) is None:

neuralforecast/common/_base_recurrent.py

@@ -11,7 +11,6 @@
 import torch
 import torch.nn as nn
 import pytorch_lightning as pl
-from pytorch_lightning.callbacks import TQDMProgressBar
 from pytorch_lightning.callbacks.early_stopping import EarlyStopping
 
 from ._scalers import TemporalNorm
@@ -127,17 +126,12 @@ def __init__(
             raise Exception("max_epochs is deprecated, use max_steps instead.")
 
         # Callbacks
-        if trainer_kwargs.get("callbacks", None) is None:
-            callbacks = [TQDMProgressBar()]
-            # Early stopping
-            if self.early_stop_patience_steps > 0:
-                callbacks += [
-                    EarlyStopping(
-                        monitor="ptl/val_loss", patience=self.early_stop_patience_steps
-                    )
-                ]
-
-            trainer_kwargs["callbacks"] = callbacks
+        if "callbacks" not in trainer_kwargs and self.early_stop_patience_steps > 0:
+            trainer_kwargs["callbacks"] = [
+                EarlyStopping(
+                    monitor="ptl/val_loss", patience=self.early_stop_patience_steps
+                )
+            ]
 
         # Add GPU accelerator if available
         if trainer_kwargs.get("accelerator", None) is None:

neuralforecast/common/_base_windows.py

@@ -11,7 +11,6 @@
 import torch
 import torch.nn as nn
 import pytorch_lightning as pl
-from pytorch_lightning.callbacks import TQDMProgressBar
 from pytorch_lightning.callbacks.early_stopping import EarlyStopping
 
 from ._scalers import TemporalNorm
@@ -137,17 +136,12 @@ def __init__(
             raise Exception("max_epochs is deprecated, use max_steps instead.")
 
         # Callbacks
-        if trainer_kwargs.get("callbacks", None) is None:
-            callbacks = [TQDMProgressBar()]
-            # Early stopping
-            if self.early_stop_patience_steps > 0:
-                callbacks += [
-                    EarlyStopping(
-                        monitor="ptl/val_loss", patience=self.early_stop_patience_steps
-                    )
-                ]
-
-            trainer_kwargs["callbacks"] = callbacks
+        if "callbacks" not in trainer_kwargs and self.early_stop_patience_steps > 0:
+            trainer_kwargs["callbacks"] = [
+                EarlyStopping(
+                    monitor="ptl/val_loss", patience=self.early_stop_patience_steps
+                )
+            ]
 
         # Add GPU accelerator if available
         if trainer_kwargs.get("accelerator", None) is None: