Merge branch 'main' into checkpoint_tutorial

Author: Svetlana Karslioglu

.github/workflows/build-tutorials.yml

@@ -34,7 +34,7 @@ jobs:
       fail-fast: false
     runs-on: ${{ matrix.runner }}
     env:
-      DOCKER_IMAGE: "308535385114.dkr.ecr.us-east-1.amazonaws.com/pytorch/pytorch-linux-bionic-cuda12.1-cudnn8-py3-gcc9"
+      DOCKER_IMAGE: "308535385114.dkr.ecr.us-east-1.amazonaws.com/pytorch/pytorch-linux-focal-cuda12.1-cudnn8-py3-gcc9"
       CUDA_VERSION: "9"
     steps:
       - name: Setup SSH (Click me for login details)
@@ -115,8 +115,9 @@ jobs:
     name: pytorch_tutorial_build_manager
     needs: worker
     runs-on: [self-hosted, linux.2xlarge]
+    environment: ${{ github.ref == 'refs/heads/main' && 'pytorchbot-env' || '' }}
     env:
-      DOCKER_IMAGE: "308535385114.dkr.ecr.us-east-1.amazonaws.com/pytorch/pytorch-linux-bionic-cuda12.1-cudnn8-py3-gcc9"
+      DOCKER_IMAGE: "308535385114.dkr.ecr.us-east-1.amazonaws.com/pytorch/pytorch-linux-focal-cuda12.1-cudnn8-py3-gcc9"
       CUDA_VERSION: "9"
     steps:
       - name: Setup SSH (Click me for login details)

_static/img/replaybuffer_traj.png

@@ -988,7 +988,7 @@ def make_replay_buffer(buffer_size, batch_size, random_crop_len, prefetch=3, prb
 # samples sub-trajectories of 2 time steps each.
 #
 # .. figure:: /_static/img/replaybuffer_traj.png
-#    :alt: Storign trajectories in the replay buffer
+#    :alt: Storing trajectories in the replay buffer
 #
 # Let's start with the number of frames stored in the buffer
 

intermediate_source/coding_ddpg.py renamed to advanced_source/coding_ddpg.py

@@ -299,7 +299,7 @@ What's new in PyTorch tutorials?
    :header: Code a DDPG Loss
    :card_description: Use TorchRL to code a DDPG Loss
    :image: _static/img/half_cheetah.gif
-   :link: intermediate/coding_ddpg.html
+   :link: advanced/coding_ddpg.html
    :tags: Reinforcement-Learning
 
 

index.rst

@@ -5,7 +5,7 @@
 **Author**: `Sean Robertson <https://github.com/spro>`_
 
 This is our second of three tutorials on "NLP From Scratch".
-In the `first tutorial </intermediate/char_rnn_classification_tutorial>`
+In the `first tutorial </intermediate/char_rnn_classification_tutorial>`_
 we used a RNN to classify names into their language of origin. This time
 we'll turn around and generate names from languages.
 

intermediate_source/char_rnn_generation_tutorial.py

@@ -2,7 +2,7 @@
 
 This directory contains tutorials and recipes demonstrating prototype features in PyTorch. 
 
-**Prototype features** are not available as part of binary distributions like PyPI or Conda (except maybe behind run-time flags). To test these features we would, depending on the feature, recommend building from master or using the nightly wheels that are made available on pytorch.org. 
+**Prototype features** are part of the release and are available as part of the binary distributions such as PyPI or Conda. To test these features you can, depending on the feature, build from the master branch or use the nightly wheels that are made available at pytorch.org. You can also by use the release wheels available from PyPI or Conda. Prototype features are provided as a technical preview and can be altered later on. The PyTorch team does not recommend using them in production pipelines.
 
 These are intentionally left out of the pytorch.org/tutorials build and will not show up on the website.