pytorch · svekars · Jul 24, 2025
diff --git a/.gitignore b/.gitignore
@@ -5,6 +5,8 @@ advanced
 pytorch_basics
 /recipes
 prototype
+/unstable
+sg_execution_times.rst
 
 #data things
 _data/

diff --git a/conf.py b/conf.py
@@ -170,9 +170,9 @@ def wrapper(*args, **kwargs):
         "intermediate_source",
         "advanced_source",
         "recipes_source",
-        "prototype_source",
+        "unstable_source",
     ],
-    "gallery_dirs": ["beginner", "intermediate", "advanced", "recipes", "prototype"],
+    "gallery_dirs": ["beginner", "intermediate", "advanced", "recipes", "unstable"],
     "filename_pattern": re.compile(SPHINX_SHOULD_RUN),
     "promote_jupyter_magic": True,
     "backreferences_dir": None,

diff --git a/index.rst b/index.rst
@@ -843,10 +843,10 @@ Additional Resources
    :maxdepth: 1
    :hidden:
 
-   recipes/recipes_index
+   recipes_index
 
 .. toctree::
    :maxdepth: 1
    :hidden:
 
-   prototype/prototype_index
+   unstable_index
diff --git a/recipes_source/recipes_index.rst → recipes_index.rst b/recipes_source/recipes_index.rst → recipes_index.rst
diff --git a/prototype_source/prototype_index.rst → unstable_index.rst b/prototype_source/prototype_index.rst → unstable_index.rst
@@ -41,119 +41,119 @@ decide if we want to upgrade the level of commitment or to fail fast.
 .. customcarditem::
    :header: (prototype) Accelerating BERT with semi-structured (2:4) sparsity
    :card_description: Prune BERT to be 2:4 sparse and accelerate for inference.
-   :image: ../_static/img/thumbnails/cropped/generic-pytorch-logo.png
-   :link: prototype/semi_structured_sparse.html
+   :image: _static/img/thumbnails/cropped/generic-pytorch-logo.png
+   :link: unstable/semi_structured_sparse.html
    :tags: Model-Optimiziation
 
 .. Modules
 
 .. customcarditem::
    :header: Skipping Module Parameter Initialization in PyTorch 1.10
    :card_description: Describes skipping parameter initialization during module construction in PyTorch 1.10, avoiding wasted computation.
-   :image: ../_static/img/thumbnails/cropped/generic-pytorch-logo.png
-   :link: ../prototype/skip_param_init.html
+   :image: _static/img/thumbnails/cropped/generic-pytorch-logo.png
+   :link: unstable/skip_param_init.html
    :tags: Modules
 
 .. vmap
 
 .. customcarditem::
    :header: Using torch.vmap
    :card_description: Learn about torch.vmap, an autovectorizer for PyTorch operations.
-   :image: ../_static/img/thumbnails/cropped/generic-pytorch-logo.png
-   :link: ../prototype/vmap_recipe.html
+   :image: _static/img/thumbnails/cropped/generic-pytorch-logo.png
+   :link: unstable/vmap_recipe.html
    :tags: vmap
 
 .. NestedTensor
 
 .. customcarditem::
    :header: Nested Tensor
    :card_description: Learn about nested tensors, the new way to batch heterogeneous-length data
-   :image: ../_static/img/thumbnails/cropped/generic-pytorch-logo.png
-   :link: ../prototype/nestedtensor.html
+   :image: _static/img/thumbnails/cropped/generic-pytorch-logo.png
+   :link: unstable/nestedtensor.html
    :tags: NestedTensor
 
 .. MaskedTensor
 
 .. customcarditem::
    :header: MaskedTensor Overview
    :card_description: Learn about masked tensors, the source of truth for specified and unspecified values
-   :image: ../_static/img/thumbnails/cropped/generic-pytorch-logo.png
-   :link: ../prototype/maskedtensor_overview.html
+   :image: _static/img/thumbnails/cropped/generic-pytorch-logo.png
+   :link: unstable/maskedtensor_overview.html
    :tags: MaskedTensor
 
 .. customcarditem::
    :header: Masked Tensor Sparsity
    :card_description: Learn about how to leverage sparse layouts (e.g. COO and CSR) in MaskedTensor
-   :image: ../_static/img/thumbnails/cropped/generic-pytorch-logo.png
-   :link: ../prototype/maskedtensor_sparsity.html
+   :image: _static/img/thumbnails/cropped/generic-pytorch-logo.png
+   :link: unstable/maskedtensor_sparsity.html
    :tags: MaskedTensor
 
 .. customcarditem::
    :header: Masked Tensor Advanced Semantics
    :card_description: Learn more about Masked Tensor's advanced semantics (reductions and comparing vs. NumPy's MaskedArray)
-   :image: ../_static/img/thumbnails/cropped/generic-pytorch-logo.png
-   :link: ../prototype/maskedtensor_advanced_semantics.html
+   :image: _static/img/thumbnails/cropped/generic-pytorch-logo.png
+   :link: unstable/maskedtensor_advanced_semantics.html
    :tags: MaskedTensor
 
 .. customcarditem::
    :header: MaskedTensor: Simplifying Adagrad Sparse Semantics
    :card_description: See a showcase on how masked tensors can enable sparse semantics and provide for a cleaner dev experience
-   :image: ../_static/img/thumbnails/cropped/generic-pytorch-logo.png
-   :link: ../prototype/maskedtensor_adagrad.html
+   :image: _static/img/thumbnails/cropped/generic-pytorch-logo.png
+   :link: unstable/maskedtensor_adagrad.html
    :tags: MaskedTensor
 
 .. Model-Optimization
 
 .. customcarditem::
    :header: Inductor Cpp Wrapper Tutorial
    :card_description: Speed up your models with Inductor Cpp Wrapper
-   :image: ../_static/img/thumbnails/cropped/generic-pytorch-logo.png
-   :link: ../prototype/inductor_cpp_wrapper_tutorial.html
+   :image: _static/img/thumbnails/cropped/generic-pytorch-logo.png
+   :link: unstable/inductor_cpp_wrapper_tutorial.html
    :tags: Model-Optimization
 
 .. customcarditem::
    :header: Inductor Windows CPU Tutorial
    :card_description: Speed up your models with Inductor On Windows CPU
-   :image: ../_static/img/thumbnails/cropped/generic-pytorch-logo.png
-   :link: ../prototype/inductor_windows.html
+   :image: _static/img/thumbnails/cropped/generic-pytorch-logo.png
+   :link: unstable/inductor_windows.html
    :tags: Model-Optimization
 
 .. customcarditem::
    :header: Use max-autotune compilation on CPU to gain additional performance boost
    :card_description: Tutorial for max-autotune mode on CPU to gain additional performance boost
-   :image: ../_static/img/thumbnails/cropped/generic-pytorch-logo.png
-   :link: ../prototype/max_autotune_on_CPU_tutorial.html
+   :image: _static/img/thumbnails/cropped/generic-pytorch-logo.png
+   :link: unstable/max_autotune_on_CPU_tutorial.html
    :tags: Model-Optimization
 
 .. Distributed
 .. customcarditem::
    :header: Flight Recorder Tutorial
    :card_description: Debug stuck jobs easily with Flight Recorder
-   :image: ../_static/img/thumbnails/cropped/generic-pytorch-logo.png
-   :link: ../prototype/flight_recorder_tutorial.html
+   :image: _static/img/thumbnails/cropped/generic-pytorch-logo.png
+   :link: unstable/flight_recorder_tutorial.html
    :tags: Distributed, Debugging, FlightRecorder
 
 .. customcarditem::
    :header: Context Parallel Tutorial
    :card_description: Parallelize the attention computation along sequence dimension
-   :image: ../_static/img/thumbnails/cropped/generic-pytorch-logo.png
-   :link: ../prototype/context_parallel.html
+   :image: _static/img/thumbnails/cropped/generic-pytorch-logo.png
+   :link: unstable/context_parallel.html
    :tags: Distributed, Context Parallel
 
 .. Integration
 .. customcarditem::
    :header: Out-of-tree extension autoloading in Python
    :card_description: Learn how to improve the seamless integration of out-of-tree extension with PyTorch based on the autoloading mechanism.
-   :image: ../_static/img/thumbnails/cropped/generic-pytorch-logo.png
-   :link: ../prototype/python_extension_autoload.html
+   :image: _static/img/thumbnails/cropped/generic-pytorch-logo.png
+   :link: unstable/python_extension_autoload.html
    :tags: Extending-PyTorch, Frontend-APIs
 
 .. GPUDirect Storage
 .. customcarditem::
    :header: (prototype) Using GPUDirect Storage
    :card_description: Learn how to use GPUDirect Storage in PyTorch.
-   :image: ../_static/img/thumbnails/cropped/generic-pytorch-logo.png
-   :link: ../prototype/gpu_direct_storage.html
+   :image: _static/img/thumbnails/cropped/generic-pytorch-logo.png
+   :link: unstable/gpu_direct_storage.html
    :tags: GPUDirect-Storage
 
 .. End of tutorial card section
@@ -166,18 +166,18 @@ decide if we want to upgrade the level of commitment or to fail fast.
    :maxdepth: 2
    :hidden:
 
-   /prototype/context_parallel
-   /prototype/flight_recorder_tutorial
-   /prototype/inductor_cpp_wrapper_tutorial
-   /prototype/inductor_windows
-   /prototype/vmap_recipe
-   /prototype/vulkan_workflow
-   /prototype/nestedtensor
-   /prototype/maskedtensor_overview
-   /prototype/maskedtensor_sparsity
-   /prototype/maskedtensor_advanced_semantics
-   /prototype/maskedtensor_adagrad
-   /prototype/python_extension_autoload
-   /prototype/gpu_direct_storage.html
-   /prototype/max_autotune_on_CPU_tutorial
-   /prototype/skip_param_init.html
+   unstable/context_parallel
+   unstable/flight_recorder_tutorial
+   unstable/inductor_cpp_wrapper_tutorial
+   unstable/inductor_windows
+   unstable/vmap_recipe
+   unstable/vulkan_workflow
+   unstable/nestedtensor
+   unstable/maskedtensor_overview
+   unstable/maskedtensor_sparsity
+   unstable/maskedtensor_advanced_semantics
+   unstable/maskedtensor_adagrad
+   unstable/python_extension_autoload
+   unstable/gpu_direct_storage.html
+   unstable/max_autotune_on_CPU_tutorial
+   unstable/skip_param_init.html
diff --git a/prototype_source/README.md → unstable_source/README.md b/prototype_source/README.md → unstable_source/README.md
diff --git a/prototype_source/README.txt → unstable_source/README.txt b/prototype_source/README.txt → unstable_source/README.txt
diff --git a/prototype_source/backend_config_tutorial.rst → unstable_source/backend_config_tutorial.rst b/prototype_source/backend_config_tutorial.rst → unstable_source/backend_config_tutorial.rst
diff --git a/prototype_source/context_parallel.rst → unstable_source/context_parallel.rst b/prototype_source/context_parallel.rst → unstable_source/context_parallel.rst
diff --git a/...type_source/distributed_rpc_profiling.rst → ...able_source/distributed_rpc_profiling.rst b/...type_source/distributed_rpc_profiling.rst → ...able_source/distributed_rpc_profiling.rst
diff --git a/...otype_source/flight_recorder_tutorial.rst → unstable_source/flight_recorder_tutorial.rst b/...otype_source/flight_recorder_tutorial.rst → unstable_source/flight_recorder_tutorial.rst
diff --git a/prototype_source/gpu_direct_storage.py → unstable_source/gpu_direct_storage.py b/prototype_source/gpu_direct_storage.py → unstable_source/gpu_direct_storage.py
diff --git a/...urce/gpu_quantization_torchao_tutorial.py → ...urce/gpu_quantization_torchao_tutorial.py b/...urce/gpu_quantization_torchao_tutorial.py → ...urce/gpu_quantization_torchao_tutorial.py
diff --git a/..._source/inductor_cpp_wrapper_tutorial.rst → ..._source/inductor_cpp_wrapper_tutorial.rst b/..._source/inductor_cpp_wrapper_tutorial.rst → ..._source/inductor_cpp_wrapper_tutorial.rst
diff --git a/prototype_source/inductor_windows.rst → unstable_source/inductor_windows.rst b/prototype_source/inductor_windows.rst → unstable_source/inductor_windows.rst
diff --git a/prototype_source/inductor_windows_cpu.rst → unstable_source/inductor_windows_cpu.rst b/prototype_source/inductor_windows_cpu.rst → unstable_source/inductor_windows_cpu.rst
diff --git a/prototype_source/ios_coreml_workflow.rst → unstable_source/ios_coreml_workflow.rst b/prototype_source/ios_coreml_workflow.rst → unstable_source/ios_coreml_workflow.rst
diff --git a/prototype_source/ios_gpu_workflow.rst → unstable_source/ios_gpu_workflow.rst b/prototype_source/ios_gpu_workflow.rst → unstable_source/ios_gpu_workflow.rst
diff --git a/prototype_source/lite_interpreter.rst → unstable_source/lite_interpreter.rst b/prototype_source/lite_interpreter.rst → unstable_source/lite_interpreter.rst
diff --git a/prototype_source/maskedtensor_adagrad.py → unstable_source/maskedtensor_adagrad.py b/prototype_source/maskedtensor_adagrad.py → unstable_source/maskedtensor_adagrad.py
diff --git a/...source/maskedtensor_advanced_semantics.py → ...source/maskedtensor_advanced_semantics.py b/...source/maskedtensor_advanced_semantics.py → ...source/maskedtensor_advanced_semantics.py
diff --git a/prototype_source/maskedtensor_overview.py → unstable_source/maskedtensor_overview.py b/prototype_source/maskedtensor_overview.py → unstable_source/maskedtensor_overview.py
diff --git a/prototype_source/maskedtensor_sparsity.py → unstable_source/maskedtensor_sparsity.py b/prototype_source/maskedtensor_sparsity.py → unstable_source/maskedtensor_sparsity.py
diff --git a/...e_source/max_autotune_on_CPU_tutorial.rst → ...e_source/max_autotune_on_CPU_tutorial.rst b/...e_source/max_autotune_on_CPU_tutorial.rst → ...e_source/max_autotune_on_CPU_tutorial.rst
diff --git a/prototype_source/nestedtensor.py → unstable_source/nestedtensor.py b/prototype_source/nestedtensor.py → unstable_source/nestedtensor.py
diff --git a/prototype_source/nnapi_mobilenetv2.rst → unstable_source/nnapi_mobilenetv2.rst b/prototype_source/nnapi_mobilenetv2.rst → unstable_source/nnapi_mobilenetv2.rst
diff --git a/prototype_source/openvino_quantizer.rst → unstable_source/openvino_quantizer.rst b/prototype_source/openvino_quantizer.rst → unstable_source/openvino_quantizer.rst
diff --git a/...type_source/python_extension_autoload.rst → ...able_source/python_extension_autoload.rst b/...type_source/python_extension_autoload.rst → ...able_source/python_extension_autoload.rst
diff --git a/prototype_source/semi_structured_sparse.rst → unstable_source/semi_structured_sparse.rst b/prototype_source/semi_structured_sparse.rst → unstable_source/semi_structured_sparse.rst
diff --git a/prototype_source/skip_param_init.rst → unstable_source/skip_param_init.rst b/prototype_source/skip_param_init.rst → unstable_source/skip_param_init.rst
diff --git a/..._source/tracing_based_selective_build.rst → ..._source/tracing_based_selective_build.rst b/..._source/tracing_based_selective_build.rst → ..._source/tracing_based_selective_build.rst
diff --git a/prototype_source/vmap_recipe.py → unstable_source/vmap_recipe.py b/prototype_source/vmap_recipe.py → unstable_source/vmap_recipe.py
diff --git a/prototype_source/vulkan_workflow.rst → unstable_source/vulkan_workflow.rst b/prototype_source/vulkan_workflow.rst → unstable_source/vulkan_workflow.rst