Merge branch 'main' into 2.6-RC-TEST

Author: svekars

advanced_source/cpp_custom_ops.rst

@@ -19,6 +19,10 @@ Custom C++ and CUDA Operators
        * PyTorch 2.4 or later
        * Basic understanding of C++ and CUDA programming
 
+.. note::
+
+  This tutorial will also work on AMD ROCm with no additional modifications.
+
 PyTorch offers a large library of operators that work on Tensors (e.g. torch.add, torch.sum, etc).
 However, you may wish to bring a new custom operator to PyTorch. This tutorial demonstrates the
 blessed path to authoring a custom operator written in C++/CUDA.

advanced_source/pendulum.py

@@ -33,9 +33,9 @@
 
   In the process, we will touch three crucial components of TorchRL:
 
-* `environments <https://pytorch.org/rl/reference/envs.html>`__
-* `transforms <https://pytorch.org/rl/reference/envs.html#transforms>`__
-* `models (policy and value function) <https://pytorch.org/rl/reference/modules.html>`__
+* `environments <https://pytorch.org/rl/stable/reference/envs.html>`__
+* `transforms <https://pytorch.org/rl/stable/reference/envs.html#transforms>`__
+* `models (policy and value function) <https://pytorch.org/rl/stable/reference/modules.html>`__
 
 """
 
@@ -384,7 +384,7 @@ def _reset(self, tensordict):
 # convenient shortcuts to the content of the output and input spec containers.
 #
 # TorchRL offers multiple :class:`~torchrl.data.TensorSpec`
-# `subclasses <https://pytorch.org/rl/reference/data.html#tensorspec>`_ to
+# `subclasses <https://pytorch.org/rl/stable/reference/data.html#tensorspec>`_ to
 # encode the environment's input and output characteristics.
 #
 # Specs shape

beginner_source/blitz/autograd_tutorial.py

@@ -191,15 +191,15 @@
 # .. math::
 #
 #
-#      J^{T}\cdot \vec{v} = m \cdot \left(\begin{array}{ccc}
+#      J^{T}\cdot \vec{v} = \left(\begin{array}{ccc}
 #       \frac{\partial y_{1}}{\partial x_{1}} & \cdots & \frac{\partial y_{m}}{\partial x_{1}}\\
 #       \vdots & \ddots & \vdots\\
 #       \frac{\partial y_{1}}{\partial x_{n}} & \cdots & \frac{\partial y_{m}}{\partial x_{n}}
 #       \end{array}\right)\left(\begin{array}{c}
 #       \frac{\partial l}{\partial y_{1}}\\
 #       \vdots\\
 #       \frac{\partial l}{\partial y_{m}}
-#       \end{array}\right) = m \cdot \left(\begin{array}{c}
+#       \end{array}\right) = \left(\begin{array}{c}
 #       \frac{\partial l}{\partial x_{1}}\\
 #       \vdots\\
 #       \frac{\partial l}{\partial x_{n}}

intermediate_source/dist_tuto.rst

@@ -47,6 +47,7 @@ the following template.
     """run.py:"""
     #!/usr/bin/env python
     import os
+    import sys
     import torch
     import torch.distributed as dist
     import torch.multiprocessing as mp
@@ -66,8 +67,12 @@ the following template.
     if __name__ == "__main__":
         world_size = 2
         processes = []
-        mp.set_start_method("spawn")
-        for rank in range(world_size):
+        if "google.colab" in sys.modules:
+            print("Running in Google Colab")
+            mp.get_context("spawn")
+        else:
+            mp.set_start_method("spawn")
+        for rank in range(size):
             p = mp.Process(target=init_process, args=(rank, world_size, run))
             p.start()
             processes.append(p)
@@ -156,7 +161,8 @@ we should not modify the sent tensor nor access the received tensor before ``req
 In other words,
 
 -  writing to ``tensor`` after ``dist.isend()`` will result in undefined behaviour.
--  reading from ``tensor`` after ``dist.irecv()`` will result in undefined behaviour.
+- reading from ``tensor`` after ``dist.irecv()`` will result in undefined
+  behaviour, until ``req.wait()`` has been executed.
 
 However, after ``req.wait()``
 has been executed we are guaranteed that the communication took place,