ci/test: use bare ipython instead of nbval (#324)

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

Author: Borda

.actions/assistant.py

@@ -113,7 +113,7 @@ def get_running_cuda_version() -> str:
         return ""
 
 
-def get_running_torch_version():
+def get_running_torch_version() -> str:
     """Extract the version of actual PyTorch for this runtime."""
     try:
         import torch
@@ -322,7 +322,13 @@ def bash_render(folder: str, output_file: str = PATH_SCRIPT_RENDER) -> Optional[
             # dry run does not execute the notebooks just takes them as they are
             cmd.append(f"cp {ipynb_file} {pub_ipynb}")
             # copy and add meta config
-            cmd += [f"cp {meta_file} {pub_meta}", f"cat {pub_meta}", f"git add {pub_meta}"]
+            cmd += [
+                f"cp {meta_file} {pub_meta}",
+                'echo "#====== START OF YAML FILE ======#"',
+                f"cat {pub_meta}",
+                'echo "#======= END OF YAML FILE =======#"',
+                f"git add {pub_meta}",
+            ]
         else:
             pip_req, pip_args = AssistantCLI._parse_requirements(folder)
             cmd += [f"pip install {pip_req} --quiet {pip_args}", "pip list"]
@@ -335,7 +341,13 @@ def bash_render(folder: str, output_file: str = PATH_SCRIPT_RENDER) -> Optional[
             # Export the actual packages used in runtime
             cmd.append(f"meta_file=$(python .actions/assistant.py update-env-details {folder})")
             # copy and add to version the enriched meta config
-            cmd += ["echo $meta_file", "cat $meta_file", "git add $meta_file"]
+            cmd += [
+                "echo $meta_file",
+                'echo "#====== START OF YAML FILE ======#"',
+                "cat $meta_file",
+                'echo "#======= END OF YAML FILE =======#"',
+                "git add $meta_file",
+            ]
         # if thumb image is linked to the notebook, copy and version it too
         if thumb_file:
             cmd += [f"cp {thumb_file} {pub_thumb}", f"git add {pub_thumb}"]
@@ -347,7 +359,7 @@ def bash_render(folder: str, output_file: str = PATH_SCRIPT_RENDER) -> Optional[
             fopen.write(os.linesep.join(cmd))
 
     @staticmethod
-    def bash_test(folder: str, output_file: str = PATH_SCRIPT_TEST) -> Optional[str]:
+    def bash_test(folder: str, output_file: str = PATH_SCRIPT_TEST, virtualenv: bool = False) -> Optional[str]:
         """Prepare bash script for running tests of a particular notebook.
 
         Args:
@@ -364,11 +376,12 @@ def bash_test(folder: str, output_file: str = PATH_SCRIPT_TEST) -> Optional[str]
 
         # prepare isolated environment with inheriting the global packages
         path_venv = os.path.join(folder, "venv")
-        cmd += [
-            f"python -m virtualenv --system-site-packages {path_venv}",
-            f"source {os.path.join(path_venv, 'bin', 'activate')}",
-            "pip --version",
-        ]
+        if virtualenv:
+            cmd += [
+                f"python -m virtualenv --system-site-packages {path_venv}",
+                f"source {os.path.join(path_venv, 'bin', 'activate')}",
+                "pip --version",
+            ]
 
         cmd.append(f"# available: {AssistantCLI.DEVICE_ACCELERATOR}")
         if AssistantCLI._valid_accelerator(folder):
@@ -378,21 +391,30 @@ def bash_test(folder: str, output_file: str = PATH_SCRIPT_TEST) -> Optional[str]
             # Export the actual packages used in runtime
             cmd.append(f"meta_file=$(python .actions/assistant.py update-env-details {folder} --base_path .)")
             # show created meta config
-            cmd += ["echo $meta_file", "cat $meta_file"]
-            cmd.append(f"python -m pytest {ipynb_file} -v --nbval --nbval-cell-timeout=300")
+            cmd += [
+                "echo $meta_file",
+                'echo "#====== START OF YAML FILE ======#"',
+                "cat $meta_file",
+                'echo "#======= END OF YAML FILE =======#"',
+            ]
+            # use standard jupyter's executable via CMD
+            cmd.append(f"jupyter execute {ipynb_file} --inplace")
         else:
             pub_ipynb = os.path.join(DIR_NOTEBOOKS, f"{folder}.ipynb")
             pub_meta = pub_ipynb.replace(".ipynb", ".yaml")
             # copy and add meta config
             cmd += [
                 f"mkdir -p {os.path.dirname(pub_meta)}",
                 f"cp {meta_file} {pub_meta}",
+                'echo "#====== START OF YAML FILE ======#"',
                 f"cat {pub_meta}",
+                'echo "#======= END OF YAML FILE =======#"',
                 f"git add {pub_meta}",
             ]
             warn("Invalid notebook's accelerator for this device. So no tests will be run!!!", RuntimeWarning)
         # deactivate and clean local environment
-        cmd += ["deactivate", f"rm -rf {os.path.join(folder, 'venv')}"]
+        if virtualenv:
+            cmd += ["deactivate", f"rm -rf {os.path.join(folder, 'venv')}"]
         if not output_file:
             return os.linesep.join(cmd)
         with open(output_file, "w") as fopen:
@@ -707,7 +729,10 @@ def update_env_details(folder: str, base_path: str = DIR_NOTEBOOKS) -> str:
 
         Args:
              folder: path to the folder
-             base_path:
+             base_path: base path with notebooks
+
+        Returns:
+            path the updated YAML file
 
         """
         meta = AssistantCLI._load_meta(folder)

.azure/ipynb-tests.yml .azure/ipynb-validate.yml.azure/ipynb-tests.yml renamed to .azure/ipynb-validate.yml

@@ -19,9 +19,12 @@ jobs:
         displayName: "Install dependencies"
 
       - bash: |
-          head=$(git rev-parse origin/main)
-          printf "Head: $head\n"
-          git diff --name-only $head --output=target-diff.txt
+          git fetch --all  # some issues with missing main :/
+          # head=$(git rev-parse origin/main)
+          # printf "Head: $head\n"  # this shall be commit hash
+          # git diff --name-only $head --output=target-diff.txt
+          git diff --name-only origin/main HEAD --output=target-diff.txt
+          cat target-diff.txt
           python .actions/assistant.py group-folders --fpath_gitdiff=target-diff.txt
           printf "Changed folders:\n"
           cat changed-folders.txt
@@ -35,7 +38,7 @@ jobs:
       - bash: echo '$(mtrx.dirs)' | python -m json.tool
         displayName: "Show matrix"
 
-  - job: nbval
+  - job: ipython
     dependsOn: check_diff
     strategy:
       matrix: $[ dependencies.check_diff.outputs['mtrx.dirs'] ]
@@ -96,4 +99,4 @@ jobs:
         env:
           KAGGLE_USERNAME: $(KAGGLE_USERNAME)
           KAGGLE_KEY: $(KAGGLE_KEY)
-        displayName: "PyTest notebook"
+        displayName: "Execute notebook"

.github/workflows/ci_test-acts.yml .github/workflows/ci_internal.yml.github/workflows/ci_test-acts.yml renamed to .github/workflows/ci_internal.yml

@@ -1,6 +1,6 @@
 # PytorchLightning Tutorials
 
-[![CI internal](https://github.com/Lightning-AI/tutorials/actions/workflows/ci_test-acts.yml/badge.svg?event=push)](https://github.com/Lightning-AI/tutorials/actions/workflows/ci_test-acts.yml)
+[![CI internal](https://github.com/Lightning-AI/tutorials/actions/workflows/ci_internal.yml/badge.svg?event=push)](https://github.com/Lightning-AI/tutorials/actions/workflows/ci_internal.yml)
 [![Build Status](https://dev.azure.com/Lightning-AI/Tutorials/_apis/build/status/Lightning-AI.tutorials%20%5Bpublish%5D?branchName=main)](https://dev.azure.com/Lightning-AI/Tutorials/_build/latest?definitionId=29&branchName=main)
 [![codecov](https://codecov.io/gh/Lightning-AI/tutorials/branch/main/graph/badge.svg?token=C6T3XOOR56)](https://codecov.io/gh/Lightning-AI/tutorials)
 [![Deploy Docs](https://github.com/Lightning-AI/tutorials/actions/workflows/docs-deploy.yml/badge.svg)](https://github.com/Lightning-AI/tutorials/actions/workflows/docs-deploy.yml)
@@ -91,7 +91,7 @@ On the back side of publishing workflow you can find in principle these three st
 # 1) convert script to notebooks
 jupytext --set-formats ipynb,py:percent notebook.py
 
-# 2) testing the created notebook
+# 2) [OPTIONAL] testing the created notebook
 pytest -v notebook.ipynb  --nbval
 
 # 3) generating notebooks outputs

README.md

@@ -2,6 +2,7 @@ setuptools>=68.0.0, <69.1.0
 matplotlib>=3.0.0, <3.9.0
 ipython[notebook]>=8.0.0, <8.17.0
 urllib3  # for ipython
+numpy <2.0 # needed for older Torch
 torch>=1.8.1, <2.1.0
 pytorch-lightning>=1.4, <2.1.0
 torchmetrics>=0.7, <1.3

_requirements/default.txt

@@ -2,5 +2,5 @@ virtualenv>=20.10
 jupytext>=1.10, <1.15  # converting
 pytest>=6.0, <7.0
 # testing with own fork with extended cell timeout
-https://github.com/Borda/nbval/archive/refs/heads/timeout-limit.zip
+# https://github.com/Borda/nbval/archive/refs/heads/timeout-limit.zip
 papermill>=2.3.4, <2.5.0  # render

_requirements/devel.txt

@@ -3,6 +3,7 @@ author: Phillip Lippe
 created: 2021-08-27
 updated: 2023-03-14
 license: CC BY-SA
+build: 1
 description: |
   This tutorial will give a short introduction to PyTorch basics, and get you setup for writing your own neural networks.
   This notebook is part of a lecture series on Deep Learning at the University of Amsterdam.

course_UvA-DL/01-introduction-to-pytorch/.meta.yml

@@ -3,7 +3,7 @@ author: Phillip Lippe
 created: 2021-07-12
 updated: 2023-03-14
 license: CC BY-SA
-build: 0
+build: 1
 tags:
   - Image
 description: |
@@ -22,7 +22,7 @@ requirements:
   - torchvision
   - matplotlib
   - tensorboard
-  - lightning>=2.0.0
+  - pytorch-lightning>=2.0.0
 accelerator:
   - CPU
   - GPU

…on-to-pytorch/Introduction_to_PyTorch.py …L/01-introduction-to-pytorch/notebook.pycourse_UvA-DL/01-introduction-to-pytorch/Introduction_to_PyTorch.py renamed to course_UvA-DL/01-introduction-to-pytorch/notebook.py course_UvA-DL/01-introduction-to-pytorch/Introduction_to_PyTorch.py renamed to course_UvA-DL/01-introduction-to-pytorch/notebook.py

@@ -9,9 +9,6 @@
 import urllib.request
 from urllib.error import HTTPError
 
-# PyTorch Lightning
-import lightning as L
-
 # Plotting
 import matplotlib
 import matplotlib.pyplot as plt
@@ -20,6 +17,9 @@
 import matplotlib_inline.backend_inline
 import numpy as np
 
+# PyTorch Lightning
+import pytorch_lightning as pl
+
 # PyTorch
 import torch
 import torch.nn as nn
@@ -28,7 +28,7 @@
 
 # Torchvision
 import torchvision
-from lightning.pytorch.callbacks import Callback, LearningRateMonitor, ModelCheckpoint
+from pytorch_lightning.callbacks import Callback, LearningRateMonitor, ModelCheckpoint
 from torchvision import transforms
 from torchvision.datasets import MNIST
 
@@ -41,7 +41,7 @@
 CHECKPOINT_PATH = os.environ.get("PATH_CHECKPOINT", "saved_models/tutorial8")
 
 # Setting the seed
-L.seed_everything(42)
+pl.seed_everything(42)
 
 # Ensure that all operations are deterministic on GPU (if used) for reproducibility
 torch.backends.cudnn.deterministic = True
@@ -465,7 +465,7 @@ def generate_samples(model, inp_imgs, steps=60, step_size=10, return_img_per_ste
 
 
 # %%
-class DeepEnergyModel(L.LightningModule):
+class DeepEnergyModel(pl.LightningModule):
     def __init__(self, img_shape, batch_size, alpha=0.1, lr=1e-4, beta1=0.0, **CNN_args):
         super().__init__()
         self.save_hyperparameters()
@@ -640,7 +640,7 @@ def on_epoch_end(self, trainer, pl_module):
 # %%
 def train_model(**kwargs):
     # Create a PyTorch Lightning trainer with the generation callback
-    trainer = L.Trainer(
+    trainer = pl.Trainer(
         default_root_dir=os.path.join(CHECKPOINT_PATH, "MNIST"),
         accelerator="auto",
         devices=1,
@@ -660,7 +660,7 @@ def train_model(**kwargs):
         print("Found pretrained model, loading...")
         model = DeepEnergyModel.load_from_checkpoint(pretrained_filename)
     else:
-        L.seed_everything(42)
+        pl.seed_everything(42)
         model = DeepEnergyModel(**kwargs)
         trainer.fit(model, train_loader, test_loader)
         model = DeepEnergyModel.load_from_checkpoint(trainer.checkpoint_callback.best_model_path)
@@ -709,7 +709,7 @@ def train_model(**kwargs):
 
 # %%
 model.to(device)
-L.seed_everything(43)
+pl.seed_everything(43)
 callback = GenerateCallback(batch_size=4, vis_steps=8, num_steps=256)
 imgs_per_step = callback.generate_imgs(model)
 imgs_per_step = imgs_per_step.cpu()
@@ -770,7 +770,7 @@ def train_model(**kwargs):
     rand_imgs = torch.rand((128,) + model.hparams.img_shape).to(model.device)
     rand_imgs = rand_imgs * 2 - 1.0
     rand_out = model.cnn(rand_imgs).mean()
-    print(f"Average score for random images: {rand_out.item():4.2f}")
+    print(f"Average score for random images: {rand_out.item()}")
 
 # %% [markdown]
 # As we hoped, the model assigns very low probability to those noisy images.
@@ -803,8 +803,8 @@ def compare_images(img1, img2):
     plt.xticks([(img1.shape[2] + 2) * (0.5 + j) for j in range(2)], labels=["Original image", "Transformed image"])
     plt.yticks([])
     plt.show()
-    print(f"Score original image: {score1:4.2f}")
-    print(f"Score transformed image: {score2:4.2f}")
+    print(f"Score original image: {score1}")
+    print(f"Score transformed image: {score2}")
 
 
 # %% [markdown]