Add a note on running in colab(#2258)

Because sphinx directives and even URLs are not working properly in the notebooks, the workaround is to use the first cell to direct users to the section on how to run tutorials in Google Colab. This PR:

- Adds a note on how to install correct version if running in Colab
- Adds `first_notebook_cell` to point to that section.
- Updates 2.0 tutorials with the requirement to install 2.0.

Author: Svetlana Karslioglu

beginner_source/colab.rst

@@ -1,5 +1,30 @@
+Running Tutorials in Google Colab
+=================================
+
+When you run a tutorial in Google Colab, there might be additional
+requirements and dependencies that you need to meet in order
+for the tutorial to work properly. This section contains notes on how to
+configure various settings in order to successfully
+run PyTorch tutorials in Google Colab.
+
+PyTorch Version in Google Colab
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+When you are running a tutorial that requires a version of PyTorch that has
+just been released, that version might not be yet available in Google Colab.
+To check that you have the required ``torch`` and compatible domain libraries
+installed, run ``!pip list``.
+
+If the installed version of PyTorch is lower than required,
+uninstall it and reinstall again by running the following commands:
+
+.. code-block:: python
+
+   !pip3 uninstall --yes torch torchaudio torchvision torchtext torchdata
+   !pip3 install torch torchaudio torchvision torchtext torchdata
+
 Using Tutorial Data from Google Drive in Colab
-==============================================
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 We've added a new feature to tutorials that allows users to open the
 notebook associated with a tutorial in Google Colab. You may need to

beginner_source/new-release-colab.rst

@@ -18,8 +18,9 @@
 # 3. Read in the CNNDM, IMDB, and Multi30k datasets and pre-process their texts in preparation for the model
 # 4. Perform text summarization, sentiment classification, and translation
 #
+# .. note::
+#    This tutorial requires PyTorch 2.0.0 or later.
 #
-
 #######################################################################
 # Data Transformation
 # -------------------
@@ -34,8 +35,7 @@
 #
 # T5 uses a SentencePiece model for text tokenization. Below, we use a pre-trained SentencePiece model to build
 # the text pre-processing pipeline using torchtext's T5Transform. Note that the transform supports both
-# batched and non-batched text input (for example, one can either pass a single sentence or a list of sentences), however
-# the T5 model expects the input to be batched.
+# batched and non-batched text input (for example, one can either pass a single sentence or a list of sentences), however the T5 model expects the input to be batched.
 #
 
 from torchtext.models import T5Transform

beginner_source/t5_tutorial.py

@@ -91,7 +91,10 @@
     'gallery_dirs': ['beginner', 'intermediate', 'advanced', 'recipes', 'prototype'],
     'filename_pattern': re.compile(SPHINX_SHOULD_RUN),
     'promote_jupyter_magic': True,
-    'backreferences_dir': None
+    'backreferences_dir': None,
+    'first_notebook_cell': ("# For tips on running notebooks in Google Colab, see\n"
+                            "# https://pytorch.org/tutorials/beginner/colab\n"
+                            "%matplotlib inline")
 }
 
 if os.getenv('GALLERY_PATTERN'):

conf.py

@@ -17,8 +17,9 @@
 
 Let's demonstrate how to do this using an ensemble of simple MLPs.
 
-.. include:: ../beginner_source/new-release-colab.rst
+.. note::
 
+   This tutorial requires PyTorch 2.0.0 or later.
 """
 
 import torch

intermediate_source/ensembling.py

@@ -12,7 +12,9 @@
 provides ways of computing various higher-order autodiff quantities
 efficiently.
 
-.. include:: ../beginner_source/new-release-colab.rst
+.. note::
+
+   This tutorial requires PyTorch 2.0.0 or later.
 
 Computing the Jacobian
 ----------------------

intermediate_source/jacobians_hessians.py

@@ -11,7 +11,9 @@
 demonstrates how to easily compute this quantity using ``torch.func``,
 composable function transforms for PyTorch.
 
-.. include:: ../beginner_source/new-release-colab.rst
+.. note::
+
+   This tutorial requires PyTorch 2.0.0 or later.
 
 Setup
 -----

intermediate_source/neural_tangent_kernels.py

@@ -10,7 +10,9 @@
 sample in a batch of data. It is a useful quantity in differential privacy,
 meta-learning, and optimization research.
 
-.. include:: ../beginner_source/new-release-colab.rst
+.. note::
+
+   This tutorial requires PyTorch 2.0.0 or later.
 
 """
 

intermediate_source/per_sample_grads.py

@@ -35,8 +35,10 @@
 # * `Memory-Efficient Attention <https://github.com/facebookresearch/xformers>`__
 # * A PyTorch implementation defined in C++
 #
-# .. literalinclude:: ../beginner_source/new-release-colab.rst
-#    :language: rst
+# .. note::
+#
+#   This tutorial requires PyTorch 2.0.0 or later.
+#
 
 import torch
 import torch.nn as nn

intermediate_source/scaled_dot_product_attention_tutorial.py

@@ -36,8 +36,6 @@
 #
 # Note: a modern NVIDIA GPU (Volta or Ampere) is recommended for this tutorial.
 #
-# .. include:: ../beginner_source/new-release-colab.rst
-#
 
 ######################################################################
 # Basic Usage