Merge branch 'main' into fix/rpc-tutorial-trajectory-mixing

Author: getchebarne

.ci/docker/requirements.txt

@@ -25,7 +25,7 @@ pydantic>=2.10
 fastapi
 matplotlib
 librosa
-torch==2.11
+torch==2.12
 torchvision
 torchdata
 networkx

.github/workflows/build-tutorials-nightly.yml

@@ -15,12 +15,12 @@ name: Build tutorials (nightly/test)
 # download the binaries in .jenkins/build.sh.
 on:
   # Only main branch for now. Uncomment the below line to enable it on PRs.
-  pull_request:
+  # pull_request:
 
   # Comment out the below line to disable on the main branch
-  push:
-    branches:
-      - main
+  # push:
+  #   branches:
+  #    - main
   workflow_dispatch:
 
 concurrency:

advanced_source/privateuseone.rst

@@ -4,7 +4,6 @@ Facilitating New Backend Integration by PrivateUse1
 In this tutorial we will walk through some necessary steps to integrate a new backend
 living outside ``pytorch/pytorch`` repo by ``PrivateUse1``. Note that this tutorial assumes that
 you already have a basic understanding of PyTorch.
-you are an advanced user of PyTorch.
 
 .. note::
 

beginner_source/blitz/tensor_tutorial.py

@@ -105,9 +105,9 @@
 #
 
 # We move our tensor to the GPU if available
-if torch.cuda.is_available():
-  tensor = tensor.to('cuda')
-  print(f"Device tensor is stored on: {tensor.device}")
+device = torch.accelerator.current_accelerator().type if torch.accelerator.is_available() else 'cpu'
+tensor = tensor.to(device)
+print(f"Device tensor is stored on: {tensor.device}")
 
 
 ######################################################################

beginner_source/ddp_series_multigpu.rst

@@ -202,6 +202,7 @@ Running the distributed training job
 Here's what the code looks like:
 
 .. code-block:: python
+
    def main(rank, world_size, total_epochs, save_every):
       ddp_setup(rank, world_size)
       dataset, model, optimizer = load_train_objs()
@@ -218,7 +219,6 @@ Here's what the code looks like:
       mp.spawn(main, args=(world_size, total_epochs, save_every,), nprocs=world_size)
 
 
-
 Further Reading
 ---------------
 

beginner_source/examples_autograd/polynomial_custom_function.py

@@ -29,19 +29,24 @@ class LegendrePolynomial3(torch.autograd.Function):
     """
 
     @staticmethod
-    def forward(ctx, input):
+    def forward(input):
         """
         In the forward pass we receive a Tensor containing the input and return
-        a Tensor containing the output. ctx is a context object that can be used
-        to stash information for backward computation. You can cache tensors for
-        use in the backward pass using the ``ctx.save_for_backward`` method. Other
-        objects can be stored directly as attributes on the ctx object, such as
-        ``ctx.my_object = my_object``. Check out `Extending torch.autograd <https://docs.pytorch.org/docs/stable/notes/extending.html#extending-torch-autograd>`_
+        a Tensor containing the output. Check out `Extending torch.autograd <https://docs.pytorch.org/docs/stable/notes/extending.html#extending-torch-autograd>`_
         for further details.
         """
-        ctx.save_for_backward(input)
         return 0.5 * (5 * input ** 3 - 3 * input)
 
+    @staticmethod
+    def setup_context(ctx, inputs, output):
+        """
+        Store input for use in the backward pass using ``ctx.save_for_backward``.
+        Other objects can be stored directly as attributes on the ctx object,
+        such as ``ctx.my_object = my_object``.
+        """
+        input, = inputs
+        ctx.save_for_backward(input)
+
     @staticmethod
     def backward(ctx, grad_output):
         """
@@ -54,8 +59,11 @@ def backward(ctx, grad_output):
 
 
 dtype = torch.float
-device = torch.device("cpu")
-# device = torch.device("cuda:0")  # Uncomment this to run on GPU
+device = (
+    torch.accelerator.current_accelerator().type
+    if torch.accelerator.is_available()
+    else "cpu"
+)
 
 # Create Tensors to hold input and outputs.
 # By default, requires_grad=False, which indicates that we do not need to

beginner_source/introyt/modelsyt_tutorial.py

@@ -48,12 +48,12 @@ class is a subclass of ``torch.Tensor``, with the special behavior that
 class TinyModel(torch.nn.Module):
 
     def __init__(self):
-        super(TinyModel, self).__init__()
+        super().__init__()
 
         self.linear1 = torch.nn.Linear(100, 200)
         self.activation = torch.nn.ReLU()
         self.linear2 = torch.nn.Linear(200, 10)
-        self.softmax = torch.nn.Softmax()
+        self.softmax = torch.nn.Softmax(dim=1)
 
     def forward(self, x):
         x = self.linear1(x)
@@ -150,7 +150,7 @@ def forward(self, x):
 class LeNet(torch.nn.Module):
 
     def __init__(self):
-        super(LeNet, self).__init__()
+        super().__init__()
         # 1 input image channel (black & white), 6 output channels, 5x5 square convolution
         # kernel
         self.conv1 = torch.nn.Conv2d(1, 6, 5)
@@ -249,7 +249,7 @@ def num_flat_features(self, x):
 class LSTMTagger(torch.nn.Module):
 
     def __init__(self, embedding_dim, hidden_dim, vocab_size, tagset_size):
-        super(LSTMTagger, self).__init__()
+        super().__init__()
         self.hidden_dim = hidden_dim
 
         self.word_embeddings = torch.nn.Embedding(vocab_size, embedding_dim)

index.rst

@@ -3,6 +3,7 @@ Welcome to PyTorch Tutorials
 
 **What's new in PyTorch tutorials?**
 
+* `Data Loading Optimization in PyTorch <https://docs.pytorch.org/tutorials/intermediate/intermediate_data_loading_tutorial.html>`__
 * `Distributed Training with Ray Train <https://docs.pytorch.org/tutorials/beginner/distributed_training_with_ray_tutorial.html>`__
 * `Serve PyTorch models at scale with Ray Serve <https://docs.pytorch.org/tutorials/beginner/serving_tutorial.html>`__
 * `Hyperparameter tuning using Ray Tune <https://docs.pytorch.org/tutorials/beginner/hyperparameter_tuning_tutorial.html>`__