Merge branch 'main' into main

Author: AnHaechan

_static/img/cat_resized.jpg

@@ -107,7 +107,7 @@ def _initialize_weights(self):
 
 # Load pretrained model weights
 model_url = 'https://s3.amazonaws.com/pytorch/test_data/export/superres_epoch100-44c6958e.pth'
-batch_size = 1    # just a random number
+batch_size = 64    # just a random number
 
 # Initialize model with the pretrained weights
 map_location = lambda storage, loc: storage
@@ -218,6 +218,32 @@ def to_numpy(tensor):
 # ONNX exporter, so please contact us in that case.
 #
 
+######################################################################
+# Timing Comparison Between Models
+# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+#
+
+######################################################################
+# Since ONNX models optimize for inference speed, running the same
+# data on an ONNX model instead of a native pytorch model should result in an 
+# improvement of up to 2x. Improvement is more pronounced with higher batch sizes.
+
+
+import time
+
+x = torch.randn(batch_size, 1, 224, 224, requires_grad=True)
+
+start = time.time()
+torch_out = torch_model(x)
+end = time.time()
+print(f"Inference of Pytorch model used {end - start} seconds")
+
+ort_inputs = {ort_session.get_inputs()[0].name: to_numpy(x)}
+start = time.time()
+ort_outs = ort_session.run(None, ort_inputs)
+end = time.time()
+print(f"Inference of ONNX model used {end - start} seconds")
+
 
 ######################################################################
 # Running the model on an image using ONNX Runtime
@@ -301,10 +327,20 @@ def to_numpy(tensor):
 # Save the image, we will compare this with the output image from mobile device
 final_img.save("./_static/img/cat_superres_with_ort.jpg")
 
+# Save resized original image (without super-resolution)
+img = transforms.Resize([img_out_y.size[0], img_out_y.size[1]])(img)
+img.save("cat_resized.jpg")
 
 ######################################################################
+# Here is the comparison between the two images: 
+#
+# .. figure:: /_static/img/cat_resized.jpg
+#
+# Low-resolution image
+# 
 # .. figure:: /_static/img/cat_superres_with_ort.jpg
-#    :alt: output\_cat
+#
+# Image after super-resolution
 #
 #
 # ONNX Runtime being a cross platform engine, you can run it across
@@ -313,7 +349,7 @@ def to_numpy(tensor):
 # ONNX Runtime can also be deployed to the cloud for model inferencing
 # using Azure Machine Learning Services. More information `here <https://docs.microsoft.com/en-us/azure/machine-learning/service/concept-onnx>`__.
 #
-# More information about ONNX Runtime's performance `here <https://github.com/microsoft/onnxruntime#high-performance>`__.
+# More information about ONNX Runtime's performance `here <https://onnxruntime.ai/docs/performance>`__.
 #
 #
 # For more information about ONNX Runtime `here <https://github.com/microsoft/onnxruntime>`__.

advanced_source/super_resolution_with_onnxruntime.py

@@ -228,18 +228,7 @@
 # integer with the ``.to()`` method. Note that ``c`` contains all the same
 # values as ``b``, but truncated to integers.
 # 
-# Available data types include:
-# 
-# -  ``torch.bool``
-# -  ``torch.int8``
-# -  ``torch.uint8``
-# -  ``torch.int16``
-# -  ``torch.int32``
-# -  ``torch.int64``
-# -  ``torch.half``
-# -  ``torch.float``
-# -  ``torch.double``
-# -  ``torch.bfloat``
+# For more information, see the `data types documentation <https://pytorch.org/docs/stable/tensor_attributes.html#torch.dtype>`__.
 # 
 # Math & Logic with PyTorch Tensors
 # ---------------------------------

beginner_source/introyt/tensors_deeper_tutorial.py

@@ -1,7 +1,8 @@
 Language Modeling with ``nn.Transformer`` and torchtext
-===========================================================
+=======================================================
 
 The content is deprecated.
 
 .. raw:: html
-   <meta http-equiv="refresh" content="0; url=https://pytorch.org/tutorials/">
+
+   <meta http-equiv="Refresh" content="1; url='https://pytorch.org/tutorials/'" />

beginner_source/transformer_tutorial.rst

@@ -1,3 +1,39 @@
+# 🎉 Docathon H1 2024 Leaderboard 🎉
+
+This is the list of the docathon contributors that have participated and contributed to the PyTorch H1 2024 docathon. 
+A big shout out to everyone who have participated! We have awarded points for each merged PR. 
+For the **easy** label, we have awarded 2 points. For the **medium** label, we have awarded 5 points. 
+For the **advanced** label, we have awarded 10 points. In some cases, we have awarded credit for the PRs that 
+were not merged or issues that have been closed without a merged PR. 
+
+| Author | Points | PR |
+|--- | --- | ---|
+| ahoblitz | 34 | https://github.com/pytorch/pytorch/pull/128566, https://github.com/pytorch/pytorch/pull/128408, https://github.com/pytorch/pytorch/pull/128171, https://github.com/pytorch/pytorch/pull/128083, https://github.com/pytorch/pytorch/pull/128082, https://github.com/pytorch/pytorch/pull/127983, https://github.com/pytorch/xla/pull/7214 | 
+| afrittoli | 25 | https://github.com/pytorch/pytorch/pull/128139, https://github.com/pytorch/pytorch/pull/128133, https://github.com/pytorch/pytorch/pull/128132, https://github.com/pytorch/pytorch/pull/128129, https://github.com/pytorch/pytorch/pull/128127 | 
+| kiszk | 20 | https://github.com/pytorch/pytorch/pull/128337, https://github.com/pytorch/pytorch/pull/128123, https://github.com/pytorch/pytorch/pull/128022, https://github.com/pytorch/pytorch/pull/128312 | 
+| loganthomas | 19 | https://github.com/pytorch/pytorch/pull/128676, https://github.com/pytorch/pytorch/pull/128192, https://github.com/pytorch/pytorch/pull/128189, https://github.com/pytorch/tutorials/pull/2922, https://github.com/pytorch/tutorials/pull/2910, https://github.com/pytorch/xla/pull/7195 | 
+| ignaciobartol | 17 | https://github.com/pytorch/pytorch/pull/128741, https://github.com/pytorch/pytorch/pull/128135, https://github.com/pytorch/pytorch/pull/127938, https://github.com/pytorch/tutorials/pull/2936 | 
+| arunppsg | 17 | https://github.com/pytorch/pytorch/pull/128391, https://github.com/pytorch/pytorch/pull/128021, https://github.com/pytorch/pytorch/pull/128018, https://github.com/pytorch-labs/torchfix/pull/59 | 
+| alperenunlu | 17 | https://github.com/pytorch/tutorials/pull/2934, https://github.com/pytorch/tutorials/pull/2909, https://github.com/pytorch/pytorch/pull/104043 | 
+| anandptl84 | 10 | https://github.com/pytorch/pytorch/pull/128196, https://github.com/pytorch/pytorch/pull/128098 | 
+| GdoongMathew | 10 | https://github.com/pytorch/pytorch/pull/128136, https://github.com/pytorch/pytorch/pull/128051 | 
+| ZhaoqiongZ | 10 | https://github.com/pytorch/pytorch/pull/127872 | 
+| ZailiWang | 10 | https://github.com/pytorch/tutorials/pull/2931 | 
+| jingxu10 | 8 | https://github.com/pytorch/pytorch/pull/127280, https://github.com/pytorch/pytorch/pull/127279, https://github.com/pytorch/pytorch/pull/127278, https://github.com/pytorch/tutorials/pull/2919 | 
+| sitamgithub-MSIT | 7 | https://github.com/pytorch/tutorials/pull/2900, https://github.com/pytorch/xla/pull/7208 | 
+| spzala | 5 | https://github.com/pytorch/pytorch/pull/128679, https://github.com/pytorch/pytorch/pull/128657 | 
+| TharinduRusira | 5 | https://github.com/pytorch/pytorch/pull/128197 | 
+| zabboud | 5 | https://github.com/pytorch/pytorch/pull/128055 | 
+| orion160 | 5 | https://github.com/pytorch/tutorials/pull/2912 | 
+| Ricktho1 | 5 | https://github.com/pytorch/xla/pull/7273 | 
+| IvanLauLinTiong | 4 | https://github.com/pytorch/pytorch/pull/128526, https://github.com/pytorch/tutorials/pull/2849 | 
+| sshkhr | 2 | https://github.com/pytorch/pytorch/pull/128155 | 
+| rk7697 | 2 | https://github.com/pytorch/pytorch/pull/127993 | 
+| hippocookie | 2 | https://github.com/pytorch/tutorials/pull/2937 | 
+| diningeachox | 2 | https://github.com/pytorch/tutorials/pull/2935 | 
+| akhil-maker | 2 | https://github.com/pytorch/tutorials/pull/2899 | 
+| saurabhkthakur | 2 | https://github.com/pytorch/tutorials/pull/2896 | 
+
 # 🎉 Docathon H2 2023 Leaderboard 🎉
 
 This is the list of the docathon contributors that have participated and contributed to the H2 2023 PyTorch docathon. 

docathon-leaderboard.md

@@ -9,6 +9,8 @@
 This tutorial shows how to use PyTorch to train a Deep Q Learning (DQN) agent
 on the CartPole-v1 task from `Gymnasium <https://gymnasium.farama.org>`__.
 
+You might find it helpful to read the original `Deep Q Learning (DQN) <https://arxiv.org/abs/1312.5602>`__ paper
+
 **Task**
 
 The agent has to decide between two actions - moving the cart left or
@@ -83,7 +85,11 @@
 plt.ion()
 
 # if GPU is to be used
-device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+device = torch.device(
+    "cuda" if torch.cuda.is_available() else
+    "mps" if torch.backends.mps.is_available() else
+    "cpu"
+)
 
 
 ######################################################################
@@ -397,7 +403,7 @@ def optimize_model():
 # can produce better results if convergence is not observed.
 #
 
-if torch.cuda.is_available():
+if torch.cuda.is_available() or torch.backends.mps.is_available():
     num_episodes = 600
 else:
     num_episodes = 50

intermediate_source/reinforcement_q_learning.py

@@ -317,6 +317,15 @@ Recipes are bite-sized, actionable examples of how to use specific PyTorch featu
    :link: ../recipes/torch_compile_user_defined_triton_kernel_tutorial.html
    :tags: Model-Optimization
 
+.. Compile Time Caching in ``torch.compile``
+
+.. customcarditem::
+   :header: Compile Time Caching in ``torch.compile``
+   :card_description: Learn how to configure compile time caching in ``torch.compile``
+   :image: ../_static/img/thumbnails/cropped/generic-pytorch-logo.png
+   :link: ../recipes/torch_compile_caching_tutorial.html
+   :tags: Model-Optimization
+
 .. Intel(R) Extension for PyTorch*
 
 .. customcarditem::

recipes_source/recipes_index.rst

@@ -0,0 +1,61 @@
+Compile Time Caching in ``torch.compile``
+=========================================================
+**Authors:** `Oguz Ulgen <https://github.com/oulgen>`_ and `Sam Larsen <https://github.com/masnesral>`_
+
+Introduction
+------------------
+
+PyTorch Inductor implements several caches to reduce compilation latency.
+This recipe demonstrates how you can configure various parts of the caching in ``torch.compile``.
+
+Prerequisites
+-------------------
+
+Before starting this recipe, make sure that you have the following:
+
+* Basic understanding of ``torch.compile``. See:
+
+  * `torch.compiler API documentation <https://pytorch.org/docs/stable/torch.compiler.html#torch-compiler>`__
+  * `Introduction to torch.compile <https://pytorch.org/tutorials/intermediate/torch_compile_tutorial.html>`__
+
+* PyTorch 2.4 or later
+
+Inductor Cache Settings
+----------------------------
+
+Most of these caches are in-memory, only used within the same process, and are transparent to the user. An exception is the FX graph cache that stores compiled FX graphs. This cache allows Inductor to avoid recompilation across process boundaries when it encounters the same graph with the same Tensor input shapes (and the same configuration). The default implementation stores compiled artifacts in the system temp directory. An optional feature also supports sharing those artifacts within a cluster by storing them in a Redis database.
+
+There are a few settings relevant to caching and to FX graph caching in particular.
+The settings are accessible via environment variables listed below or can be hard-coded in Inductor’s config file.
+
+TORCHINDUCTOR_FX_GRAPH_CACHE
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+This setting enables the local FX graph cache feature, i.e., by storing artifacts in the host’s temp directory. ``1`` enables, and any other value disables it. By default, the disk location is per username, but users can enable sharing across usernames by specifying ``TORCHINDUCTOR_CACHE_DIR`` (below).
+
+TORCHINDUCTOR_CACHE_DIR
+~~~~~~~~~~~~~~~~~~~~~~~~
+This setting specifies the location of all on-disk caches. By default, the location is in the system temp directory under ``torchinductor_<username>``, for example, ``/tmp/torchinductor_myusername``.
+
+Note that if ``TRITON_CACHE_DIR`` is not set in the environment, Inductor sets the Triton cache directory to this same temp location, under the Triton subdirectory.
+
+TORCHINDUCTOR_FX_GRAPH_REMOTE_CACHE
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+This setting enables the remote FX graph cache feature. The current implementation uses Redis. ``1`` enables caching, and any other value disables it. The following environment variables configure the host and port of the Redis server:
+
+``TORCHINDUCTOR_REDIS_HOST`` (defaults to ``localhost``)
+``TORCHINDUCTOR_REDIS_PORT`` (defaults to ``6379``)
+
+Note that if Inductor locates a remote cache entry, it stores the compiled artifact in the local on-disk cache; that local artifact would be served on subsequent runs on the same machine.
+
+TORCHINDUCTOR_AUTOTUNE_REMOTE_CACHE
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+This setting enables a remote cache for Inductor’s autotuner. As with the remote FX graph cache, the current implementation uses Redis. ``1`` enables caching, and any other value disables it. The same host / port environment variables listed above apply to this cache.
+
+TORCHINDUCTOR_FORCE_DISABLE_CACHES
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+Set this value to ``1`` to disable all Inductor caching. This setting is useful for tasks like experimenting with cold-start compile times or forcing recompilation for debugging purposes.
+
+Conclusion
+-------------
+In this recipe, we have learned that PyTorch Inductor's caching mechanisms significantly reduce compilation latency by utilizing both local and remote caches, which operate seamlessly in the background without requiring user intervention.
+Additionally, we explored the various settings and environment variables that allow users to configure and optimize these caching features according to their specific needs.