replace TensorBoard by CSV logger in PL examples

lightning_examples/augmentation_kornia/augmentation.py

@@ -9,6 +9,7 @@
 import torch.nn as nn
 import torchmetrics
 import torchvision
+from IPython.core.display import display
 from kornia import image_to_tensor, tensor_to_image
 from kornia.augmentation import ColorJitter, RandomChannelShuffle, RandomHorizontalFlip, RandomThinPlateSpline
 from pytorch_lightning import LightningModule, Trainer
@@ -190,7 +191,7 @@ def val_dataloader(self):
  accelerator="auto",
  devices=1 if torch.cuda.is_available() else None, # limiting got iPython runs
  max_epochs=10,
- logger=CSVLogger(save_dir="logs/", name="cifar10-resnet18"),
+ logger=CSVLogger(save_dir="logs/"),
 )
 
 # Train the model ⚡
@@ -203,15 +204,5 @@ def val_dataloader(self):
 metrics = pd.read_csv(f"{trainer.logger.log_dir}/metrics.csv")
 del metrics["step"]
 metrics.set_index("epoch", inplace=True)
-print(metrics.dropna(axis=1, how="all").head())
-g = sn.relplot(data=metrics, kind="line")
-plt.gcf().set_size_inches(12, 4)
-plt.grid()
-
-# %% [markdown]
-# ## Tensorboard
-
-# %%
-# Start tensorboard.
-# # %load_ext tensorboard
-# # %tensorboard --logdir lightning_logs/
+display(metrics.dropna(axis=1, how="all").head())
+sn.relplot(data=metrics, kind="line")

lightning_examples/basic-gan/gan.py

@@ -159,7 +159,7 @@ def __init__(
  b1: float = 0.5,
  b2: float = 0.999,
  batch_size: int = BATCH_SIZE,
- **kwargs
+ **kwargs,
  ):
  super().__init__()
  self.save_hyperparameters()

lightning_examples/cifar10-baseline/.meta.yml

@@ -11,5 +11,7 @@ description: >
 requirements:
  - torchvision
  - lightning-bolts
+ - pandas
+ - seaborn
 accelerator:
  - GPU

lightning_examples/cifar10-baseline/baseline.py

@@ -5,16 +5,19 @@
 # %%
 import os
 
+import pandas as pd
+import seaborn as sn
 import torch
 import torch.nn as nn
 import torch.nn.functional as F
 import torchvision
+from IPython.core.display import display
 from pl_bolts.datamodules import CIFAR10DataModule
 from pl_bolts.transforms.dataset_normalizations import cifar10_normalization
 from pytorch_lightning import LightningModule, Trainer, seed_everything
 from pytorch_lightning.callbacks import LearningRateMonitor
 from pytorch_lightning.callbacks.progress import TQDMProgressBar
-from pytorch_lightning.loggers import TensorBoardLogger
+from pytorch_lightning.loggers import CSVLogger
 from torch.optim.lr_scheduler import OneCycleLR
 from torch.optim.swa_utils import AveragedModel, update_bn
 from torchmetrics.functional import accuracy
@@ -142,13 +145,21 @@ def configure_optimizers(self):
  max_epochs=30,
  accelerator="auto",
  devices=1 if torch.cuda.is_available() else None, # limiting got iPython runs
- logger=TensorBoardLogger("lightning_logs/", name="resnet"),
+ logger=CSVLogger(save_dir="logs/"),
  callbacks=[LearningRateMonitor(logging_interval="step"), TQDMProgressBar(refresh_rate=10)],
 )
 
 trainer.fit(model, cifar10_dm)
 trainer.test(model, datamodule=cifar10_dm)
 
+# %%
+
+metrics = pd.read_csv(f"{trainer.logger.log_dir}/metrics.csv")
+del metrics["step"]
+metrics.set_index("epoch", inplace=True)
+display(metrics.dropna(axis=1, how="all").head())
+sn.relplot(data=metrics, kind="line")
+
 # %% [markdown]
 # ### Bonus: Use [Stochastic Weight Averaging](https://arxiv.org/abs/1803.05407) to get a boost on performance
 #
@@ -199,14 +210,17 @@ def on_train_end(self):
  max_epochs=20,
  accelerator="auto",
  devices=1 if torch.cuda.is_available() else None, # limiting got iPython runs
- logger=TensorBoardLogger("lightning_logs/", name="swa_resnet"),
  callbacks=[TQDMProgressBar(refresh_rate=20)],
+ logger=CSVLogger(save_dir="logs/"),
 )
 
 swa_trainer.fit(swa_model, cifar10_dm)
 swa_trainer.test(swa_model, datamodule=cifar10_dm)
 
 # %%
-# Start tensorboard.
-# %reload_ext tensorboard
-# %tensorboard --logdir lightning_logs/
+
+metrics = pd.read_csv(f"{trainer.logger.log_dir}/metrics.csv")
+del metrics["step"]
+metrics.set_index("epoch", inplace=True)
+display(metrics.dropna(axis=1, how="all").head())
+sn.relplot(data=metrics, kind="line")

lightning_examples/mnist-hello-world/.meta.yml

@@ -10,6 +10,8 @@ description: In this notebook, we'll go over the basics of lightning by preparin
  models to train on the [MNIST Handwritten Digits dataset](https://en.wikipedia.org/wiki/MNIST_database).
 requirements:
  - torchvision
+ - pandas
+ - seaborn
 accelerator:
  - CPU
  - GPU

lightning_examples/mnist-hello-world/hello-world.py

@@ -1,9 +1,13 @@
 # %%
 import os
 
+import pandas as pd
+import seaborn as sn
 import torch
+from IPython.core.display import display
 from pytorch_lightning import LightningModule, Trainer
 from pytorch_lightning.callbacks.progress import TQDMProgressBar
+from pytorch_lightning.loggers import CSVLogger
 from torch import nn
 from torch.nn import functional as F
 from torch.utils.data import DataLoader, random_split
@@ -204,6 +208,7 @@ def test_dataloader(self):
  devices=1 if torch.cuda.is_available() else None, # limiting got iPython runs
  max_epochs=3,
  callbacks=[TQDMProgressBar(refresh_rate=20)],
+ logger=CSVLogger(save_dir="logs/"),
 )
 trainer.fit(model)
 
@@ -230,6 +235,9 @@ def test_dataloader(self):
 # In Colab, you can use the TensorBoard magic function to view the logs that Lightning has created for you!
 
 # %%
-# Start tensorboard.
-# %load_ext tensorboard
-# %tensorboard --logdir lightning_logs/
+
+metrics = pd.read_csv(f"{trainer.logger.log_dir}/metrics.csv")
+del metrics["step"]
+metrics.set_index("epoch", inplace=True)
+display(metrics.dropna(axis=1, how="all").head())
+sn.relplot(data=metrics, kind="line")

lightning_examples/reinforce-learning-DQN/.meta.yml

@@ -15,6 +15,8 @@ description: |
 requirements:
  - gym
  - pygame
+ - pandas
+ - seaborn
 accelerator:
  - CPU
  - GPU

lightning_examples/reinforce-learning-DQN/dqn.py

@@ -5,8 +5,12 @@
 
 import gym
 import numpy as np
+import pandas as pd
+import seaborn as sn
 import torch
+from IPython.core.display import display
 from pytorch_lightning import LightningModule, Trainer
+from pytorch_lightning.loggers import CSVLogger
 from torch import Tensor, nn
 from torch.optim import Adam, Optimizer
 from torch.utils.data import DataLoader
@@ -364,11 +368,15 @@ def get_device(self, batch) -> str:
  devices=1 if torch.cuda.is_available() else None, # limiting got iPython runs
  max_epochs=150,
  val_check_interval=50,
+ logger=CSVLogger(save_dir="logs/"),
 )
 
 trainer.fit(model)
 
 # %%
-# Start tensorboard.
-# %load_ext tensorboard
-# %tensorboard --logdir lightning_logs/
+
+metrics = pd.read_csv(f"{trainer.logger.log_dir}/metrics.csv")
+del metrics["step"]
+metrics.set_index("epoch", inplace=True)
+display(metrics.dropna(axis=1, how="all").head())
+sn.relplot(data=metrics, kind="line")

templates/img-classify/classify.py

@@ -6,6 +6,7 @@
 import pandas as pd
 import seaborn as sn
 from flash.image import ImageClassificationData, ImageClassifier
+from IPython.core.display import display
 from pytorch_lightning.loggers import CSVLogger
 
 PATH_DATASETS = os.environ.get("PATH_DATASETS", ".")
@@ -41,7 +42,7 @@
 metrics = pd.read_csv(f"{trainer.logger.log_dir}/metrics.csv")
 del metrics["step"]
 metrics.set_index("epoch", inplace=True)
-print(metrics.dropna(axis=1, how="all").head())
+display(metrics.dropna(axis=1, how="all").head())
 
 g = sn.relplot(data=metrics, kind="line")
 plt.gcf().set_size_inches(12, 4)