Merge branch 'main' into malfet/add-pyspelling

Author: malfet

.circleci/README.md

@@ -0,0 +1 @@
+Do not edit `config.yml` directly, make all the changes to `config.yml.in` and then run `regenerate.py`

.circleci/config.yml

@@ -190,6 +190,7 @@ jobs:
     resource_class: gpu.nvidia.small.multi
   pytorch_tutorial_pr_build_worker_1:
     <<: *pytorch_tutorial_build_worker_defaults
+    resource_class: gpu.nvidia.medium
   pytorch_tutorial_pr_build_worker_10:
     <<: *pytorch_tutorial_build_worker_defaults
   pytorch_tutorial_pr_build_worker_11:
@@ -234,6 +235,7 @@ jobs:
     resource_class: gpu.nvidia.small.multi
   pytorch_tutorial_trunk_build_worker_1:
     <<: *pytorch_tutorial_build_worker_defaults
+    resource_class: gpu.nvidia.medium
   pytorch_tutorial_trunk_build_worker_10:
     <<: *pytorch_tutorial_build_worker_defaults
   pytorch_tutorial_trunk_build_worker_11:

.circleci/regenerate.py

@@ -26,16 +26,20 @@ def indent(indentation, data_list):
 def jobs(pr_or_trunk, num_workers=20, indentation=2):
     jobs = {}
 
-    # all tutorials that need gpu.nvidia.small.multi machines will be routed
-    # by get_files_to_run.py to 0th worker
+    # all tutorials that need gpu.nvidia.small.multi machines will be routed by
+    # get_files_to_run.py to 0th worker, similarly for gpu.nvidia.medium and the
+    # 1st worker
     needs_gpu_nvidia_small_multi = [0]
+    needs_gpu_nvidia_medium = [1]
     jobs[f"pytorch_tutorial_{pr_or_trunk}_build_manager"] = {
         "<<": "*pytorch_tutorial_build_manager_defaults"
     }
     for i in range(num_workers):
         job_info = {"<<": "*pytorch_tutorial_build_worker_defaults"}
         if i in needs_gpu_nvidia_small_multi:
             job_info["resource_class"] = "gpu.nvidia.small.multi"
+        if i in needs_gpu_nvidia_medium:
+            job_info["resource_class"] = "gpu.nvidia.medium"
         jobs[f"pytorch_tutorial_{pr_or_trunk}_build_worker_{i}"] = job_info
 
     return indent(indentation, jobs).replace("'", "")

.github/ISSUE_TEMPLATE/bug-report.yml

@@ -0,0 +1,60 @@
+name: 🐛 Bug Report
+description: Create a tutorial bug report
+title: "[BUG] - <title>"
+labels: [
+  "bug"
+]
+
+body:
+- type: markdown
+  attributes:
+    value: >
+      #### Before submitting a bug, please make sure the issue hasn't been already addressed by searching through [the existing and past issues](https://github.com/pytorch/tutorials/issues?q=is%3Aissue+sort%3Acreated-desc+).
+- type: textarea
+  attributes:
+    label: Add Link
+    description: | 
+      **Add the link to the tutorial***
+    placeholder: |
+      Link to the tutorial on the website:
+  validations:
+    required: true 
+- type: textarea
+  attributes:
+    label: Describe the bug
+    description: | 
+      **Add the bug description**      
+    placeholder: |
+      Provide a detailed description of the issue with code samples if relevant
+      ```python
+
+      # Sample code to reproduce the problem if relevant
+      ```
+
+      **Expected Result:** (Describe what you were expecting to see)
+
+
+      **Actual Result:** (Describe the result)
+
+      ```
+      The error message you got, with the full traceback.
+      ```
+
+  validations:
+    required: true
+- type: textarea
+  attributes:
+    label: Describe your environment
+    description: |
+      **Describe the environment you encountered the bug in:**
+    placeholder: |
+      * Platform (i.e macOS, Linux, Google Colab):
+      * CUDA (yes/no, version?):
+      * PyTorch version (run `python -c "import torch; print(torch.__version__)"`):
+
+  validations:
+    required: true
+- type: markdown
+  attributes:
+    value: >
+      Thanks for contributing 🎉!

.github/ISSUE_TEMPLATE/feature-request.yml

@@ -0,0 +1,37 @@
+name: 🚀 Feature request
+description: Submit a proposal for a new PyTorch tutorial or improvement of an existing tutorial
+title: "💡 [REQUEST] - <title>"
+labels: [
+  "feature"
+]
+
+body:
+- type: textarea
+  attributes:
+    label: 🚀 Descirbe the improvement or the new tutorial
+    description: |
+      **Describe the improvement**
+    placeholder: |
+      Explain why this improvement or new tutorial is important. For example, *"This tutorial will help users to better understand feature X of PyTorch."* If there is a tutorial that you propose to replace, add here. If this is related to another GitHub issue, add a link here.
+  validations:
+    required: true
+- type: textarea
+  attributes:
+    label: Existing tutorials on this topic
+    description: | 
+      **Add a list of existing tutorials on the same topic.**
+    placeholder: |
+      List tutorials that already explain this functionality if exist. On pytorch.org or elsewhere.
+      * Link
+      * Link
+- type: textarea
+  attributes:
+    label: Additional context
+    description: |
+      **Add additional context**
+    placeholder: |
+      Add any other context or screenshots about the feature request.
+- type: markdown
+  attributes:
+    value: >
+      Thanks for contributing 🎉!

.jenkins/get_files_to_run.py

@@ -39,15 +39,23 @@ def add_to_shard(i, filename):
             shard_jobs,
         )
 
+    all_other_files = all_files.copy()
     needs_gpu_nvidia_small_multi = list(
         filter(lambda x: get_needs_machine(x) == "gpu.nvidia.small.multi", all_files,)
     )
+    needs_gpu_nvidia_medium = list(
+        filter(lambda x: get_needs_machine(x) == "gpu.nvidia.medium", all_files,)
+    )
     for filename in needs_gpu_nvidia_small_multi:
         # currently, the only job that uses gpu.nvidia.small.multi is the 0th worker,
         # so we'll add all the jobs that need this machine to the 0th worker
         add_to_shard(0, filename)
-
-    all_other_files = [x for x in all_files if x not in needs_gpu_nvidia_small_multi]
+        all_other_files.remove(filename)
+    for filename in needs_gpu_nvidia_medium:
+        # currently, the only job that uses gpu.nvidia.medium is the 1st worker,
+        # so we'll add all the jobs that need this machine to the 1st worker
+        add_to_shard(1, filename)
+        all_other_files.remove(filename)
 
     sorted_files = sorted(all_other_files, key=get_duration, reverse=True,)
 

.jenkins/get_sphinx_filenames.py

@@ -0,0 +1,13 @@
+from pathlib import Path
+from typing import List
+
+from get_files_to_run import get_all_files
+from validate_tutorials_built import NOT_RUN
+
+
+def get_files_for_sphinx() -> List[str]:
+    all_py_files = get_all_files()
+    return [x for x in all_py_files if all(y not in x for y in NOT_RUN)]
+
+
+SPHINX_SHOULD_RUN = "|".join(get_files_for_sphinx())

.jenkins/metadata.json

@@ -24,5 +24,8 @@
   },
   "intermediate_source/model_parallel_tutorial.py": {
     "needs": "gpu.nvidia.small.multi"
+  },
+  "intermediate_source/torch_compile_tutorial.py": {
+    "needs": "gpu.nvidia.medium"
   }
 }

.jenkins/validate_tutorials_built.py

@@ -9,51 +9,49 @@
 # the file name to explain why, like intro.html), or fix the tutorial and remove it from this list).
 
 NOT_RUN = [
-    "basics/intro",  # no code
-    "translation_transformer",
-    "profiler",
-    "saving_loading_models",
-    "introyt/captumyt",
-    "introyt/trainingyt",
-    "examples_nn/polynomial_module",
-    "examples_nn/dynamic_net",
-    "examples_nn/polynomial_optim",
-    "former_torchies/autograd_tutorial_old",
-    "former_torchies/tensor_tutorial_old",
-    "examples_autograd/polynomial_autograd",
-    "examples_autograd/polynomial_custom_function",
-    "parametrizations",
-    "mnist_train_nas",  # used by ax_multiobjective_nas_tutorial.py
-    "fx_conv_bn_fuser",
-    "super_resolution_with_onnxruntime",
-    "ddp_pipeline",  # requires 4 gpus
-    "fx_graph_mode_ptq_dynamic",
-    "vmap_recipe",
-    "torchscript_freezing",
-    "nestedtensor",
-    "recipes/saving_and_loading_models_for_inference",
-    "recipes/saving_multiple_models_in_one_file",
-    "recipes/loading_data_recipe",
-    "recipes/tensorboard_with_pytorch",
-    "recipes/what_is_state_dict",
-    "recipes/profiler_recipe",
-    "recipes/save_load_across_devices",
-    "recipes/warmstarting_model_using_parameters_from_a_different_model",
-    "torch_compile_tutorial_",
-    "recipes/dynamic_quantization",
-    "recipes/saving_and_loading_a_general_checkpoint",
-    "recipes/benchmark",
-    "recipes/tuning_guide",
-    "recipes/zeroing_out_gradients",
-    "recipes/defining_a_neural_network",
-    "recipes/timer_quick_start",
-    "recipes/amp_recipe",
-    "recipes/Captum_Recipe",
-    "flask_rest_api_tutorial",
-    "text_to_speech_with_torchaudio",
+    "beginner_source/basics/intro",  # no code
+    "beginner_source/translation_transformer",
+    "beginner_source/profiler",
+    "beginner_source/saving_loading_models",
+    "beginner_source/introyt/captumyt",
+    "beginner_source/examples_nn/polynomial_module",
+    "beginner_source/examples_nn/dynamic_net",
+    "beginner_source/examples_nn/polynomial_optim",
+    "beginner_source/former_torchies/autograd_tutorial_old",
+    "beginner_source/former_torchies/tensor_tutorial_old",
+    "beginner_source/examples_autograd/polynomial_autograd",
+    "beginner_source/examples_autograd/polynomial_custom_function",
+    "intermediate_source/parametrizations",
+    "intermediate_source/mnist_train_nas",  # used by ax_multiobjective_nas_tutorial.py
+    "intermediate_source/fx_conv_bn_fuser",
+    "advanced_source/super_resolution_with_onnxruntime",
+    "advanced_source/ddp_pipeline",  # requires 4 gpus
+    "prototype_source/fx_graph_mode_ptq_dynamic",
+    "prototype_source/vmap_recipe",
+    "prototype_source/torchscript_freezing",
+    "prototype_source/nestedtensor",
+    "recipes_source/recipes/saving_and_loading_models_for_inference",
+    "recipes_source/recipes/saving_multiple_models_in_one_file",
+    "recipes_source/recipes/loading_data_recipe",
+    "recipes_source/recipes/tensorboard_with_pytorch",
+    "recipes_source/recipes/what_is_state_dict",
+    "recipes_source/recipes/profiler_recipe",
+    "recipes_source/recipes/save_load_across_devices",
+    "recipes_source/recipes/warmstarting_model_using_parameters_from_a_different_model",
+    "recipes_source/recipes/dynamic_quantization",
+    "recipes_source/recipes/saving_and_loading_a_general_checkpoint",
+    "recipes_source/recipes/benchmark",
+    "recipes_source/recipes/tuning_guide",
+    "recipes_source/recipes/zeroing_out_gradients",
+    "recipes_source/recipes/defining_a_neural_network",
+    "recipes_source/recipes/timer_quick_start",
+    "recipes_source/recipes/amp_recipe",
+    "recipes_source/recipes/Captum_Recipe",
+    "intermediate_source/flask_rest_api_tutorial",
+    "intermediate_source/text_to_speech_with_torchaudio",
+    "intermediate_source/tensorboard_profiler_tutorial" # reenable after 2.0 release.
 ]
 
-
 def tutorial_source_dirs() -> List[Path]:
     return [
         p.relative_to(REPO_ROOT).with_name(p.stem[:-7])
@@ -68,6 +66,7 @@ def main() -> None:
         glob_path = f"{tutorial_source_dir}/**/*.html"
         html_file_paths += docs_dir.glob(glob_path)
 
+    should_not_run = [f'{x.replace("_source", "")}.html' for x in NOT_RUN]
     did_not_run = []
     for html_file_path in html_file_paths:
         with open(html_file_path, "r", encoding="utf-8") as html_file:
@@ -78,9 +77,7 @@ def main() -> None:
             if (
                 "Total running time of the script: ( 0 minutes  0.000 seconds)"
                 in elem.text
-                and not any(
-                    html_file_path.match(file) for file in NOT_RUN
-                )
+                and not any(html_file_path.match(file) for file in should_not_run)
             ):
                 did_not_run.append(html_file_path.as_posix())
 

_static/img/invpendulum.gif

@@ -458,7 +458,8 @@ quantizing for x86 architectures. This configuration does the following:
     per_channel_quantized_model = load_model(saved_model_dir + float_model_file)  
     per_channel_quantized_model.eval()  
     per_channel_quantized_model.fuse_model()  
-    per_channel_quantized_model.qconfig = torch.ao.quantization.get_default_qconfig('fbgemm')
+    # The old 'fbgemm' is still available but 'x86' is the recommended default.
+    per_channel_quantized_model.qconfig = torch.ao.quantization.get_default_qconfig('x86')
     print(per_channel_quantized_model.qconfig)  
 
     torch.ao.quantization.prepare(per_channel_quantized_model, inplace=True)
@@ -534,8 +535,9 @@ We fuse modules as before
     qat_model = load_model(saved_model_dir + float_model_file)  
     qat_model.fuse_model()  
 
-    optimizer = torch.optim.SGD(qat_model.parameters(), lr = 0.0001)  
-    qat_model.qconfig = torch.ao.quantization.get_default_qat_qconfig('fbgemm')
+    optimizer = torch.optim.SGD(qat_model.parameters(), lr = 0.0001) 
+    # The old 'fbgemm' is still available but 'x86' is the recommended default. 
+    qat_model.qconfig = torch.ao.quantization.get_default_qat_qconfig('x86')
   
 Finally, ``prepare_qat`` performs the "fake quantization", preparing the model for quantization-aware training