Merge branch 'master' into quant-fix

Author: andrewor14

.circleci/config.yml

@@ -173,7 +173,7 @@ pytorch_tutorial_build_worker_defaults: &pytorch_tutorial_build_worker_defaults
   environment:
     DOCKER_IMAGE: "308535385114.dkr.ecr.us-east-1.amazonaws.com/pytorch/pytorch-linux-xenial-cuda10.2-cudnn7-py3-gcc7"
     CUDA_VERSION: "9"
-  resource_class: gpu.medium
+  resource_class: gpu.nvidia.small
   <<: *pytorch_tutorial_build_defaults
 
 pytorch_tutorial_build_manager_defaults: &pytorch_tutorial_build_manager_defaults
@@ -223,6 +223,7 @@ jobs:
 
   pytorch_tutorial_pr_build_worker_3:
     <<: *pytorch_tutorial_build_worker_defaults
+    resource_class: gpu.nvidia.small.multi
 
   pytorch_tutorial_pr_build_worker_4:
     <<: *pytorch_tutorial_build_worker_defaults
@@ -253,12 +254,14 @@ jobs:
 
   pytorch_tutorial_pr_build_worker_13:
     <<: *pytorch_tutorial_build_worker_defaults
+    resource_class: gpu.nvidia.small.multi
 
   pytorch_tutorial_pr_build_worker_14:
     <<: *pytorch_tutorial_build_worker_defaults
 
   pytorch_tutorial_pr_build_worker_15:
     <<: *pytorch_tutorial_build_worker_defaults
+    resource_class: gpu.nvidia.medium
 
   pytorch_tutorial_pr_build_worker_16:
     <<: *pytorch_tutorial_build_worker_defaults

.devcontainer/requirements.txt

@@ -24,7 +24,7 @@ ipython
 # to run examples
 pandas
 scikit-image
-pillow==8.3.2
+pillow==9.0.0
 wget
 
 # for codespaces env

_static/img/thumbnails/cropped/Customize-Process-Group-Backends-Using-Cpp-Extensions.png

@@ -160,7 +160,7 @@ def __getitem__(self, idx):
 
 
 def __init__(self, annotations_file, img_dir, transform=None, target_transform=None):
-    self.img_labels = pd.read_csv(annotations_file, names=['file_name', 'label'])
+    self.img_labels = pd.read_csv(annotations_file)
     self.img_dir = img_dir
     self.transform = transform
     self.target_transform = target_transform

_static/img/thumbnails/cropped/realtime_rpi.png

@@ -26,8 +26,7 @@
 from torch import nn
 from torch.utils.data import DataLoader
 from torchvision import datasets
-from torchvision.transforms import ToTensor, Lambda, Compose
-import matplotlib.pyplot as plt
+from torchvision.transforms import ToTensor
 
 ######################################################################
 # PyTorch offers domain-specific libraries such as `TorchText <https://pytorch.org/text/stable/index.html>`_,
@@ -67,8 +66,8 @@
 test_dataloader = DataLoader(test_data, batch_size=batch_size)
 
 for X, y in test_dataloader:
-    print("Shape of X [N, C, H, W]: ", X.shape)
-    print("Shape of y: ", y.shape, y.dtype)
+    print(f"Shape of X [N, C, H, W]: {X.shape}")
+    print(f"Shape of y: {y.shape} {y.dtype}")
     break
 
 ######################################################################

beginner_source/basics/data_tutorial.py

@@ -112,7 +112,7 @@
 
 # We move our tensor to the GPU if available
 if torch.cuda.is_available():
-    tensor = tensor.to('cuda')
+    tensor = tensor.to("cuda")
 
 
 ######################################################################
@@ -124,9 +124,9 @@
 # **Standard numpy-like indexing and slicing:**
 
 tensor = torch.ones(4, 4)
-print('First row: ', tensor[0])
-print('First column: ', tensor[:, 0])
-print('Last column:', tensor[..., -1])
+print(f"First row: {tensor[0]}")
+print(f"First column: {tensor[:, 0]}")
+print(f"Last column: {tensor[..., -1]}")
 tensor[:,1] = 0
 print(tensor)
 
@@ -172,7 +172,7 @@
 # Operations that store the result into the operand are called in-place. They are denoted by a ``_`` suffix.
 # For example: ``x.copy_(y)``, ``x.t_()``, will change ``x``.
 
-print(tensor, "\n")
+print(f"{tensor} \n")
 tensor.add_(5)
 print(tensor)
 

beginner_source/basics/quickstart_tutorial.py

@@ -39,6 +39,7 @@
 import torch.nn as nn
 import torch.optim as optim
 from torch.optim import lr_scheduler
+import torch.backends.cudnn as cudnn
 import numpy as np
 import torchvision
 from torchvision import datasets, models, transforms
@@ -47,6 +48,7 @@
 import os
 import copy
 
+cudnn.benchmark = True
 plt.ion()   # interactive mode
 
 ######################################################################

beginner_source/basics/tensorqs_tutorial.py

@@ -55,7 +55,7 @@ Welcome to PyTorch Tutorials
    :image: _static/img/thumbnails/cropped/60-min-blitz.png
    :link: beginner/basics/intro.html
    :tags: Getting-Started
-   
+
 .. customcarditem::
    :header: Introduction to PyTorch on YouTube
    :card_description: An introduction to building a complete ML workflow with PyTorch. Follows the PyTorch Beginner Series on YouTube.
@@ -120,7 +120,7 @@ Welcome to PyTorch Tutorials
    :image: _static/img/thumbnails/cropped/DCGAN-Tutorial.png
    :link: beginner/dcgan_faces_tutorial.html
    :tags: Image/Video
-   
+
 .. customcarditem::
    :header: Spatial Transformer Networks Tutorial
    :card_description: Learn how to augment your network using a visual attention mechanism.
@@ -352,6 +352,13 @@ Welcome to PyTorch Tutorials
    :link: advanced/torch-script-parallelism.html
    :tags: Frontend-APIs,TorchScript,C++
 
+.. customcarditem::
+   :header: Real Time Inference on Raspberry Pi 4
+   :card_description: This tutorial covers how to run quantized and fused models on a Raspberry Pi 4 at 30 fps.
+   :image: _static/img/thumbnails/cropped/realtime_rpi.png
+   :link: intermediate/realtime_rpi.html
+   :tags: TorchScript,Model Optimization,Image/Video,Quantization
+
 .. customcarditem::
    :header: Autograd in C++ Frontend
    :card_description: The autograd package helps build flexible and dynamic nerural netorks. In this tutorial, exploreseveral examples of doing autograd in PyTorch C++ frontend
@@ -496,6 +503,13 @@ Welcome to PyTorch Tutorials
    :link: intermediate/dist_tuto.html
    :tags: Parallel-and-Distributed-Training
 
+.. customcarditem::
+   :header: Customize Process Group Backends Using Cpp Extensions
+   :card_description: Extend ProcessGroup with custom collective communication implementations.
+   :image: _static/img/thumbnails/cropped/Customize-Process-Group-Backends-Using-Cpp-Extensions.png
+   :link: intermediate/process_group_cpp_extension_tutorial.html
+   :tags: Parallel-and-Distributed-Training
+
 .. customcarditem::
    :header: Getting Started with Distributed RPC Framework
    :card_description: Learn how to build distributed training using the torch.distributed.rpc package.
@@ -646,7 +660,7 @@ Additional Resources
    beginner/basics/autogradqs_tutorial
    beginner/basics/optimization_tutorial
    beginner/basics/saveloadrun_tutorial
-   
+
 .. toctree::
    :maxdepth: 2
    :hidden:
@@ -736,6 +750,7 @@ Additional Resources
    beginner/Intro_to_TorchScript_tutorial
    advanced/cpp_export
    advanced/super_resolution_with_onnxruntime
+   intermediate/realtime_rpi
 
 .. toctree::
    :maxdepth: 2
@@ -799,6 +814,7 @@ Additional Resources
    intermediate/model_parallel_tutorial
    intermediate/ddp_tutorial
    intermediate/dist_tuto
+   intermediate/process_group_cpp_extension_tutorial
    intermediate/rpc_tutorial
    intermediate/rpc_param_server_tutorial
    intermediate/dist_pipeline_parallel_tutorial

beginner_source/transfer_learning_tutorial.py

@@ -218,7 +218,7 @@ def summary(self, should_sort : bool = False) -> str:
 #   https://github.com/pytorch/pytorch/issues/51393
 # * BatchNorm2d also takes up significant time. We can continue this
 #   line of thinking and optimize this in the Conv-BN Fusion with FX
-#   tutorial TODO: link
+#   `tutorial <https://pytorch.org/tutorials/intermediate/fx_conv_bn_fuser.html>`_. 
 #
 #
 # Conclusion