Merge branch 'master' into suraj813-patch-3

Author: Svetlana Karslioglu

.devcontainer/requirements.txt

@@ -24,7 +24,7 @@ ipython
 # to run examples
 pandas
 scikit-image
-pillow==9.0.1
+pillow==9.3.0
 wget
 
 # for codespaces env

.jenkins/validate_tutorials_built.py

@@ -39,7 +39,7 @@
     "recipes/profiler_recipe",
     "recipes/save_load_across_devices",
     "recipes/warmstarting_model_using_parameters_from_a_different_model",
-    "dynamo_tutorial",
+    "torch_compile_tutorial_",
     "recipes/dynamic_quantization",
     "recipes/saving_and_loading_a_general_checkpoint",
     "recipes/benchmark",

_static/img/reinforcement_learning_diagram.drawio

@@ -0,0 +1 @@
+<mxfile host="app.diagrams.net" modified="2022-10-01T16:00:40.980Z" agent="5.0 (X11)" etag="_qbqVrrm3wUvm_i0-Q9T" version="20.4.0" type="device"><diagram id="aSXDm0BvLjt-Za0vl2Tv" name="Page-1">5Vpbc+MmFP41nmkfmpGEpMiPjTftzrTZZtbbbbYvHSxhiRQJFeHb/vqChG4gx95ElqfTeCaGwwEO37lwDskMLNL9zwzmyQONEJk5VrSfgXczx7GtuSW+JOVQUXzXrggxw5FiaglL/BXVMxV1gyNU9Bg5pYTjvE8MaZahkPdokDG667OtKenvmsMYGYRlCIlJ/QNHPKmogWe19PcIx0m9s22pkRTWzIpQJDCiuw4J3M/AglHKq1a6XyAiwatxqeb9dGS0EYyhjJ8z4flQfCient3PwfKXVfTn40P6/vMPapUtJBt14I8oJ/AgaA8opeygZOeHGhBGN1mE5JrWDNztEszRMoehHN0JExC0hKdE9GzRXGNCFpRQVs4Fa09+BD1mMMJC7s7YqvyIsYIz+jfqjPjljxhRsiLG0f4oCHYDrbBJRFPE5SEsNcGt1aPMMVDdXatbu1ZY0tGrr2hQmVPcrNwiLhoK9G9QgGMo4Lec41T6gWN9535v4C/WFLaPTmM/AlrebR8t2z0TruBScAEDLgaziKaCtoI8TAy4wg3bltYqIUFZ9KOMCaIbElgUOOxD1rftai0UGQFCg0/sRzcsRKfdjEMWI37KGkx1dOD2BtCuaQwRyPG2L+6QCtQOjxSLgxz1DRdoWqyOqWZ1I42+kNdfCOjmUOFgLCR0U0aemi2XDMVxgQ3ztK0X5fJtjR/0+EWjkqA1z0YHr7dYz7DYR0pwKM/5AfFRw2sEUbAOh8PrLYI+sgbDaxig1foy4dWxrh1fAzOACodeqi5lPKExzSC5b6laGGh5fqU0V8g/I84PKluBG06HQu8okcN/W+Q4OyS8CWTfsPFPlbz/Cxu/eg5hm0nEmPcg2mP+JNs3nup96Yy823c7h/HvTu8/cXfqV9H8lXen5xxJuUa+O91A2yd4+e709LvTP8Hvvsh/mbvWNtPDUR0hE+I9NVNF50vrFrLb+kLZq52hcaCO+9hTuY9zpvt413Qf3Vqc29e6z1xbyJkm9TSKb51fS4mdKVJP2zXc4fc8grwsLb3rlpaODog3cItaU96ijUdNmirWscHuXq03jjdVeLDnZ8aHI+qcJsGspewFdp8Iee8ivJU7Ehxn5YD/z0a+qN0RtOZtT7Ri9Q1Tac3ZqsjLvvWJQZzhLBbNUmtqXSFnuXQzb5zd7Bvxa5FQWkgvbB4vJDxCgXVlCEOOaXZhURwpylJQiRQFZdsL7wfkfh9RSFkkGql6XrQ2KiRddG9X7t2+rF10L6/ElUpu5VZ/ZWUt1D/piuk76/K8pWyq5S+lHiVi23oGaA9E7PlAxG4Yxw/ZZr4X1q5Vu9AE6V8wP5UAyt4jYlgcG7HrlUhVGL1WkgeO5EDf/r5oDdcuo9dIeqUPXk7ygK/xn3iPNACxJkgKHTMpHNVJBmod6+Z2snzmqmWMrlCgVx/nWjjQLc+7jIUDvYw5ZeFA43emsFjzCf0iYd2ava6q7z2LTVbX18XdyaDvX9UjNIMBevl2tkdo71VATyrG8ghd4LcV6qLb/oW/Ym//TwLc/ws=</diagram></mxfile>

_static/img/reinforcement_learning_diagram.jpg

@@ -34,7 +34,7 @@
 ------------------
 You can run this tutorial in a couple of ways:
 
-- **In the cloud**: This is the easiest way to get started! Each section has a "Run in Microsoft Learn" link at the top, which opens an integrated notebook in Microsoft Learn with the code in a fully-hosted environment.
+- **In the cloud**: This is the easiest way to get started! Each section has a "Run in Microsoft Learn" and "Run in Google Colab" link at the top, which opens an integrated notebook in Microsoft Learn or Google Colab, respectively, with the code in a fully-hosted environment.
 - **Locally**: This option requires you to setup PyTorch and TorchVision first on your local machine (`installation instructions <https://pytorch.org/get-started/locally/>`_). Download the notebook or copy the code into your favorite IDE.
 
 

beginner_source/basics/intro.py

@@ -126,7 +126,7 @@ DDP materials are listed below:
    described in the
    `Single-Machine Model Parallel Best Practices <../intermediate/model_parallel_tutorial.html>`__
    tutorial.
-3. The `Launching and configuring distributed data parallel applications <https://github.com/pytorch/examples/blob/stable/distributed/ddp/README.md>`__
+3. The `Launching and configuring distributed data parallel applications <https://github.com/pytorch/examples/blob/main/distributed/ddp/README.md>`__
    document shows how to use the DDP launching script.
 4. The `Shard Optimizer States With ZeroRedundancyOptimizer <../recipes/zero_redundancy_optimizer.html>`__
    recipe demonstrates how `ZeroRedundancyOptimizer <https://pytorch.org/docs/stable/distributed.optim.html>`__

beginner_source/dist_overview.rst

@@ -3,8 +3,8 @@
 ======================================================
 
 This tutorial shows:
-    - How to train a translation model from scratch using Transformer. 
-    - Use tochtext library to access  `Multi30k <http://www.statmt.org/wmt16/multimodal-task.html#task1>`__ dataset to train a German to English translation model.
+    - How to train a translation model from scratch using Transformer.
+    - Use torchtext library to access  `Multi30k <http://www.statmt.org/wmt16/multimodal-task.html#task1>`__ dataset to train a German to English translation model.
 """
 
 
@@ -14,12 +14,12 @@
 #
 # `torchtext library <https://pytorch.org/text/stable/>`__ has utilities for creating datasets that can be easily
 # iterated through for the purposes of creating a language translation
-# model. In this example, we show how to use torchtext's inbuilt datasets, 
+# model. In this example, we show how to use torchtext's inbuilt datasets,
 # tokenize a raw text sentence, build vocabulary, and numericalize tokens into tensor. We will use
 # `Multi30k dataset from torchtext library <https://pytorch.org/text/stable/datasets.html#multi30k>`__
-# that yields a pair of source-target raw sentences. 
+# that yields a pair of source-target raw sentences.
 #
-# To access torchtext datasets, please install torchdata following instructions at https://github.com/pytorch/data. 
+# To access torchtext datasets, please install torchdata following instructions at https://github.com/pytorch/data.
 #
 
 from torchtext.data.utils import get_tokenizer
@@ -61,18 +61,18 @@ def yield_tokens(data_iter: Iterable, language: str) -> List[str]:
 UNK_IDX, PAD_IDX, BOS_IDX, EOS_IDX = 0, 1, 2, 3
 # Make sure the tokens are in order of their indices to properly insert them in vocab
 special_symbols = ['<unk>', '<pad>', '<bos>', '<eos>']
- 
+
 for ln in [SRC_LANGUAGE, TGT_LANGUAGE]:
-    # Training data Iterator 
+    # Training data Iterator
     train_iter = Multi30k(split='train', language_pair=(SRC_LANGUAGE, TGT_LANGUAGE))
-    # Create torchtext's Vocab object 
+    # Create torchtext's Vocab object
     vocab_transform[ln] = build_vocab_from_iterator(yield_tokens(train_iter, ln),
                                                     min_freq=1,
                                                     specials=special_symbols,
                                                     special_first=True)
 
-# Set UNK_IDX as the default index. This index is returned when the token is not found. 
-# If not set, it throws RuntimeError when the queried token is not found in the Vocabulary. 
+# Set UNK_IDX as the default index. This index is returned when the token is not found.
+# If not set, it throws RuntimeError when the queried token is not found in the Vocabulary.
 for ln in [SRC_LANGUAGE, TGT_LANGUAGE]:
   vocab_transform[ln].set_default_index(UNK_IDX)
 
@@ -82,14 +82,14 @@ def yield_tokens(data_iter: Iterable, language: str) -> List[str]:
 #
 # Transformer is a Seq2Seq model introduced in `“Attention is all you
 # need” <https://papers.nips.cc/paper/2017/file/3f5ee243547dee91fbd053c1c4a845aa-Paper.pdf>`__
-# paper for solving machine translation tasks. 
+# paper for solving machine translation tasks.
 # Below, we will create a Seq2Seq network that uses Transformer. The network
 # consists of three parts. First part is the embedding layer. This layer converts tensor of input indices
 # into corresponding tensor of input embeddings. These embedding are further augmented with positional
-# encodings to provide position information of input tokens to the model. The second part is the 
-# actual `Transformer <https://pytorch.org/docs/stable/generated/torch.nn.Transformer.html>`__ model. 
-# Finally, the output of Transformer model is passed through linear layer
-# that give un-normalized probabilities for each token in the target language. 
+# encodings to provide position information of input tokens to the model. The second part is the
+# actual `Transformer <https://pytorch.org/docs/stable/generated/torch.nn.Transformer.html>`__ model.
+# Finally, the output of the Transformer model is passed through linear layer
+# that gives un-normalized probabilities for each token in the target language.
 #
 
 
@@ -130,7 +130,7 @@ def __init__(self, vocab_size: int, emb_size):
     def forward(self, tokens: Tensor):
         return self.embedding(tokens.long()) * math.sqrt(self.emb_size)
 
-# Seq2Seq Network 
+# Seq2Seq Network
 class Seq2SeqTransformer(nn.Module):
     def __init__(self,
                  num_encoder_layers: int,
@@ -164,7 +164,7 @@ def forward(self,
                 memory_key_padding_mask: Tensor):
         src_emb = self.positional_encoding(self.src_tok_emb(src))
         tgt_emb = self.positional_encoding(self.tgt_tok_emb(trg))
-        outs = self.transformer(src_emb, tgt_emb, src_mask, tgt_mask, None, 
+        outs = self.transformer(src_emb, tgt_emb, src_mask, tgt_mask, None,
                                 src_padding_mask, tgt_padding_mask, memory_key_padding_mask)
         return self.generator(outs)
 
@@ -179,9 +179,9 @@ def decode(self, tgt: Tensor, memory: Tensor, tgt_mask: Tensor):
 
 
 ######################################################################
-# During training, we need a subsequent word mask that will prevent model to look into
+# During training, we need a subsequent word mask that will prevent the model from looking into
 # the future words when making predictions. We will also need masks to hide
-# source and target padding tokens. Below, let's define a function that will take care of both. 
+# source and target padding tokens. Below, let's define a function that will take care of both.
 #
 
 
@@ -204,7 +204,7 @@ def create_mask(src, tgt):
 
 
 ######################################################################
-# Let's now define the parameters of our model and instantiate the same. Below, we also 
+# Let's now define the parameters of our model and instantiate the same. Below, we also
 # define our loss function which is the cross-entropy loss and the optmizer used for training.
 #
 torch.manual_seed(0)
@@ -218,7 +218,7 @@ def create_mask(src, tgt):
 NUM_ENCODER_LAYERS = 3
 NUM_DECODER_LAYERS = 3
 
-transformer = Seq2SeqTransformer(NUM_ENCODER_LAYERS, NUM_DECODER_LAYERS, EMB_SIZE, 
+transformer = Seq2SeqTransformer(NUM_ENCODER_LAYERS, NUM_DECODER_LAYERS, EMB_SIZE,
                                  NHEAD, SRC_VOCAB_SIZE, TGT_VOCAB_SIZE, FFN_HID_DIM)
 
 for p in transformer.parameters():
@@ -234,11 +234,11 @@ def create_mask(src, tgt):
 ######################################################################
 # Collation
 # ---------
-#   
-# As seen in the ``Data Sourcing and Processing`` section, our data iterator yields a pair of raw strings. 
-# We need to convert these string pairs into the batched tensors that can be processed by our ``Seq2Seq`` network 
-# defined previously. Below we define our collate function that convert batch of raw strings into batch tensors that
-# can be fed directly into our model.   
+#
+# As seen in the ``Data Sourcing and Processing`` section, our data iterator yields a pair of raw strings.
+# We need to convert these string pairs into the batched tensors that can be processed by our ``Seq2Seq`` network
+# defined previously. Below we define our collate function that converts a batch of raw strings into batch tensors that
+# can be fed directly into our model.
 #
 
 
@@ -254,8 +254,8 @@ def func(txt_input):
 
 # function to add BOS/EOS and create tensor for input sequence indices
 def tensor_transform(token_ids: List[int]):
-    return torch.cat((torch.tensor([BOS_IDX]), 
-                      torch.tensor(token_ids), 
+    return torch.cat((torch.tensor([BOS_IDX]),
+                      torch.tensor(token_ids),
                       torch.tensor([EOS_IDX])))
 
 # src and tgt language text transforms to convert raw strings into tensors indices
@@ -276,9 +276,9 @@ def collate_fn(batch):
     src_batch = pad_sequence(src_batch, padding_value=PAD_IDX)
     tgt_batch = pad_sequence(tgt_batch, padding_value=PAD_IDX)
     return src_batch, tgt_batch
-    
+
 ######################################################################
-# Let's define training and evaluation loop that will be called for each 
+# Let's define training and evaluation loop that will be called for each
 # epoch.
 #
 
@@ -289,7 +289,7 @@ def train_epoch(model, optimizer):
     losses = 0
     train_iter = Multi30k(split='train', language_pair=(SRC_LANGUAGE, TGT_LANGUAGE))
     train_dataloader = DataLoader(train_iter, batch_size=BATCH_SIZE, collate_fn=collate_fn)
-    
+
     for src, tgt in train_dataloader:
         src = src.to(DEVICE)
         tgt = tgt.to(DEVICE)
@@ -328,7 +328,7 @@ def evaluate(model):
         src_mask, tgt_mask, src_padding_mask, tgt_padding_mask = create_mask(src, tgt_input)
 
         logits = model(src, tgt_input, src_mask, tgt_mask,src_padding_mask, tgt_padding_mask, src_padding_mask)
-        
+
         tgt_out = tgt[1:, :]
         loss = loss_fn(logits.reshape(-1, logits.shape[-1]), tgt_out.reshape(-1))
         losses += loss.item()
@@ -350,7 +350,7 @@ def evaluate(model):
     print((f"Epoch: {epoch}, Train loss: {train_loss:.3f}, Val loss: {val_loss:.3f}, "f"Epoch time = {(end_time - start_time):.3f}s"))
 
 
-# function to generate output sequence using greedy algorithm 
+# function to generate output sequence using greedy algorithm
 def greedy_decode(model, src, src_mask, max_len, start_symbol):
     src = src.to(DEVICE)
     src_mask = src_mask.to(DEVICE)

beginner_source/translation_transformer.py

@@ -519,10 +519,10 @@ What's new in PyTorch tutorials?
    :tags: Model-Optimization,Best-Practice,Ax,TorchX
 
 .. customcarditem::
-   :header: TorchDynamo and TorchInductor Tutorial
-   :card_description: Speed up your models with minimal code changes using TorchDynamo and TorchInductor, the latest PyTorch compiler solution.
+   :header: torch.compile Tutorial
+   :card_description: Speed up your models with minimal code changes using torch.compile, the latest PyTorch compiler solution.
    :image: _static/img/thumbnails/cropped/generic-pytorch-logo.png
-   :link: intermediate/dynamo_tutorial.html
+   :link: intermediate/torch_compile_tutorial.html
    :tags: Model-Optimization
 
 .. Parallel-and-Distributed-Training
@@ -909,7 +909,7 @@ Additional Resources
    intermediate/torchserve_with_ipex_2
    intermediate/nvfuser_intro_tutorial
    intermediate/ax_multiobjective_nas_tutorial
-   intermediate/dynamo_tutorial
+   intermediate/torch_compile_tutorial
 
 .. toctree::
    :maxdepth: 2