Update

.ci/docker/requirements-docs.txt

@@ -1,23 +1,24 @@
-sphinx==5.3.0
+sphinx==6.2.1
 #Description: This is used to generate PyTorch docs
-#Pinned versions: 5.3.0
+#Pinned versions: 7.2.6
+
 -e git+https://github.com/pytorch/pytorch_sphinx_theme.git@1657ad2fc1acdc98aa719eebecbb0128a7c13ce4#egg=pytorch_sphinx_theme2
 
+sphinx-remove-toctrees==1.0.0.post1
+#Description: This is used to generate PyTorch docs
+#Pinned versions: 1.0.0.post1
+
 # TODO: sphinxcontrib.katex 0.9.0 adds a local KaTeX server to speed up pre-rendering
 # but it doesn't seem to work and hangs around idly. The initial thought that it is probably
 # something related to Docker setup. We can investigate this later.
 
-sphinxcontrib.katex==0.8.6
+sphinxcontrib.katex==0.9.10
 #Description: This is used to generate PyTorch docs
-#Pinned versions: 0.8.6
+#Pinned versions: 0.9.10
 
-sphinxext-opengraph==0.9.1
+sphinx_sitemap==2.7.1
-#Description: This is used to generate PyTorch docs
-#Pinned versions: 0.9.1
-
-sphinx_sitemap==2.6.0
 #Description: This is used to generate sitemap for PyTorch docs
-#Pinned versions: 2.6.0
+#Pinned versions: 2.7.1
 
 matplotlib==3.5.3 ; python_version < "3.13"
 matplotlib==3.6.3 ; python_version >= "3.13"
@@ -29,17 +30,17 @@ tensorboard==2.18.0 ; python_version >= "3.13"
 #Description: This is used to generate PyTorch docs
 #Pinned versions: 2.13.0
 
-breathe==4.34.0
+breathe==4.35.0
 #Description: This is used to generate PyTorch C++ docs
-#Pinned versions: 4.34.0
+#Pinned versions: 4.35.0
 
-exhale==0.2.3
+exhale==0.3.7
 #Description: This is used to generate PyTorch C++ docs
-#Pinned versions: 0.2.3
+#Pinned versions: 0.3.7
 
-docutils==0.16
+docutils==0.18.1
 #Description: This is used to generate PyTorch C++ docs
-#Pinned versions: 0.16
+#Pinned versions: 0.18.1
 
 bs4==0.0.1
 #Description: This is used to generate PyTorch C++ docs
@@ -49,13 +50,24 @@ IPython==8.12.0
 #Description: This is used to generate PyTorch functorch docs
 #Pinned versions: 8.12.0
 
+<<<<<<< HEAD
 myst-nb==0.17.2
 #Description: This is used to generate PyTorch functorch and torch.compile docs.
 #Pinned versions: 0.17.2
+=======
+myst-nb==1.2.0
+#Description: This is used to generate PyTorch functorch docs
+#Pinned versions: 1.2.0
+>>>>>>> 195382ce28e (Update)
 
 # The following are required to build torch.distributed.elastic.rendezvous.etcd* docs
 python-etcd==0.4.5
-sphinx-copybutton==0.5.0
+sphinx-copybutton==0.5.2
-sphinx-design==0.4.0
+sphinx-design==0.6.1
 sphinxcontrib-mermaid==1.0.0
+<<<<<<< HEAD
 myst-parser==0.18.1
+=======
+myst-parser==3.0.1
+myst-nb
+>>>>>>> 195382ce28e (Update)

.ci/pytorch/python_doc_push_script.sh

@@ -83,6 +83,10 @@ rm -rf pytorch || true
 pushd "$pt_checkout"
 pushd docs
 
+# Profile the docs build to see what is taking the longest
+python -m cProfile -o docs_build.prof -m sphinx.cmd.build -b html -d build/doctrees source build/html
+python -c "import pstats; p = pstats.Stats('docs_build.prof'); p.sort_stats('cumtime').print_stats(50)"
+
 # Build the docs
 if [ "$is_main_doc" = true ]; then
   build_docs html || exit $?

.github/workflows/_docs.yml

@@ -74,7 +74,11 @@ jobs:
           - docs_type: python
             runner: ${{ inputs.runner_prefix }}linux.2xlarge
             # It takes less than 30m to finish python docs unless there are issues
-            timeout-minutes: 30
+            timeout-minutes: 60
+          - docs_type: functorch
+            runner: ${{ inputs.runner_prefix }}linux.2xlarge
+            # It takes less than 15m to finish functorch docs unless there are issues
+            timeout-minutes: 15
     # Set a fixed name for this job instead of using the current matrix-generated name, i.e. build-docs (cpp, linux.12xlarge, 180)
     # The current name requires updating the database last docs push query from test-infra every time the matrix is updated
     name: build-docs-${{ matrix.docs_type }}-${{ inputs.push }}

docs/source/_templates/autosummary/function.rst

@@ -0,0 +1,7 @@
+.. role:: hidden
+    :class: hidden-section
+.. currentmodule:: {{ module }}
+
+{{ name | underline }}
+
+.. autofunction:: {{ fullname }}

docs/source/_templates/autosummary/method.rst

@@ -0,0 +1,7 @@
+.. role:: hidden
+    :class: hidden-section
+.. currentmodule:: {{ module }}
+
+{{ name | underline }}
+
+.. automethod:: {{ fullname }}

docs/source/conf.py

@@ -66,6 +66,7 @@ extensions = [
     "sphinx.ext.linkcode",
     "sphinxcontrib.mermaid",
     "sphinx_sitemap",
+    "sphinx_remove_toctrees"
 ]
 
 myst_enable_extensions = [
@@ -74,6 +75,9 @@ myst_enable_extensions = [
     "html_image",
 ]
 
+# Remove the "generated" tag from the toctree to allow for faster builds
+remove_from_toctrees = ["generated/*"]
+
 html_baseurl = "https://docs.pytorch.org/docs/stable/"  # needed for sphinx-sitemap
 sitemap_locales = [None]
 sitemap_excludes = [
@@ -93,8 +97,10 @@ numpydoc_show_class_members = False
 # autosectionlabel throws warnings if section names are duplicated.
 # The following tells autosectionlabel to not throw a warning for
 # duplicated section names that are in different documents.
+
 autosectionlabel_prefix_document = True
 
+
 # katex options
 #
 #
@@ -207,6 +213,41 @@ templates_path = [
 ]
 # TODO: document these and remove them from here.
 
+autosummary_filename_map = {
+    'torch.nn.utils.prune.identity': 'torch.nn.utils.prune.identity_function',
+    'torch.nn.utils.prune.Identity': 'torch.nn.utils.prune.Identity_class',
+    'torch.optim.adamw.adamw': 'torch.optim.adamw.adamw_function',
+    'torch.optim.adamw.AdamW': 'torch.optim.adamw.AdamW_class',
+    'torch.optim.asgd.asgd': 'torch.optim.asgd.asgd_function',
+    'torch.optim.asgd.ASGD': 'torch.optim.asgd.ASGD_class',
+    'torch.optim.nadam.nadam': 'torch.optim.nadam.nadam_function',
+    'torch.optim.nadam.NAdam': 'torch.optim.nadam.NAdam_class',
+    'torch.optim.radam.radam': 'torch.optim.radam.radam_function',
+    'torch.optim.radam.RAdam': 'torch.optim.radam.RAdam_class',
+    'torch.optim.rmsprop.rmsprop': 'torch.optim.rmsprop.rmsprop_function',
+    'torch.optim.rmsprop.RMSprop': 'torch.optim.rmsprop.RMSprop_class',
+    'torch.optim.rprop.rprop': 'torch.optim.rprop.rprop_function',
+    'torch.optim.rprop.Rprop': 'torch.optim.rprop.Rprop_class',
+    'torch.optim.sgd.sgd': 'torch.optim.sgd.sgd_function',
+    'torch.optim.sgd.SGD': 'torch.optim.sgd.SGD_class',
+    'torch.optim.adadelta.adadelta': 'torch.optim.adadelta.adadelta_function',
+    'torch.optim.adadelta.Adadelta': 'torch.optim.adadelta.Adadelta_class',
+    'torch.optim.adagrad.adagrad': 'torch.optim.adagrad.adagrad_function',
+    'torch.optim.adagrad.Adagrad': 'torch.optim.adagrad.Adagrad_class',
+    'torch.optim.adam.adam': 'torch.optim.adam.adam_function',
+    'torch.optim.adam.Adam': 'torch.optim.adam.Adam_class',
+    'torch.optim.adamax.adamax': 'torch.optim.adamax.adamax_function',
+    'torch.optim.adamax.Adamax': 'torch.optim.adamax.Adamax_class',
+    'torch.mtia.stream': 'torch.mtia.stream_function',
+    'torch.mtia.Stream': 'torch.mtia.Stream_class',
+    'torch.cpu.stream': 'torch.cpu.stream_function',
+    'torch.cpu.Stream': 'torch.cpu.Stream_class',
+    'torch.cuda.stream': 'torch.cuda.stream_function',
+    'torch.cuda.Stream': 'torch.cuda.Stream_class',
+    'torch.xpu.stream': 'torch.xpu.stream_function',
+    'torch.xpu.Stream': 'torch.xpu.Stream_class',
+}
+
 coverage_ignore_functions = [
     # torch
     "typename",

docs/source/deterministic.md

@@ -21,10 +21,10 @@
     The following operations will fill uninitialized memory when this setting is
     turned on:
 
-        * :func:`torch.Tensor.resize_` when called with a tensor that is not
+        * :meth:`torch.Tensor.resize_` when called with a tensor that is not
           quantized
         * :func:`torch.empty`
         * :func:`torch.empty_strided`
         * :func:`torch.empty_permuted`
         * :func:`torch.empty_like`
-```
+```

torch/__init__.py

@@ -1409,7 +1409,7 @@ def use_deterministic_algorithms(
         * :func:`torch.nn.functional.grid_sample` when attempting to differentiate a CUDA tensor
         * :func:`torch.cumsum` when called on a CUDA tensor when dtype is floating point or complex
         * :func:`torch.Tensor.scatter_reduce` when ``reduce='prod'`` and called on CUDA tensor
-        * :func:`torch.Tensor.resize_` when called with a quantized tensor
+        * :meth:`torch.Tensor.resize_` when called with a quantized tensor
 
     In addition, several operations fill uninitialized memory when this setting
     is turned on and when

torch/_torch_docs.py

@@ -14181,7 +14181,7 @@ are freshly created instead of aliasing the input.
 add_docstr(
     torch.expand_copy,
     r"""
-Performs the same operation as :func:`torch.Tensor.expand`, but all output tensors
+Performs the same operation as :meth:`torch.Tensor.expand`, but all output tensors
 are freshly created instead of aliasing the input.
 """,
 )

torch/autograd/gradcheck.py

@@ -1999,7 +1999,7 @@ def gradcheck(
     .. warning::
        If any checked tensor in :attr:`input` has overlapping memory, i.e.,
        different indices pointing to the same memory address (e.g., from
-       :func:`torch.Tensor.expand`), this check will likely fail because the numerical
+       :meth:`torch.Tensor.expand`), this check will likely fail because the numerical
        gradients computed by point perturbation at such indices will change
        values at all other indices that share the same memory address.
 
@@ -2153,7 +2153,7 @@ def gradgradcheck(
     .. warning::
        If any checked tensor in :attr:`input` and :attr:`grad_outputs` has
        overlapping memory, i.e., different indices pointing to the same memory
-       address (e.g., from :func:`torch.Tensor.expand`), this check will likely fail
+       address (e.g., from :meth:`torch.Tensor.expand`), this check will likely fail
        because the numerical gradients computed by point perturbation at such
        indices will change values at all other indices that share the same
        memory address.