Merge branch 'main' into patch-1

Author: svekars

.ci/docker/requirements.txt

@@ -13,7 +13,7 @@ tqdm==4.66.1
 numpy==1.24.4
 matplotlib
 librosa
-torch
+torch==2.3
 torchvision
 torchtext
 torchdata
@@ -38,6 +38,8 @@ torchmultimodal-nightly # needs to be updated to stable as soon as it's avaialab
 onnx
 onnxscript
 onnxruntime
+evaluate
+accelerate>=0.20.1
 
 importlib-metadata==6.8.0
 

.jenkins/build.sh

@@ -23,8 +23,7 @@ sudo apt-get install -y pandoc
 # Nightly - pip install --pre torch torchvision torchaudio -f https://download.pytorch.org/whl/nightly/cu102/torch_nightly.html
 # Install 2.2 for testing - uncomment to install nightly binaries (update the version as needed).
 # pip uninstall -y torch torchvision torchaudio torchtext torchdata
-# pip3 install torch==2.2.0 torchvision torchaudio --no-cache-dir --index-url https://download.pytorch.org/whl/test/cu121
-# pip3 install torchdata torchtext --index-url https://download.pytorch.org/whl/test/cpu
+# pip3 install torch==2.3.0 torchvision torchaudio --no-cache-dir --index-url https://download.pytorch.org/whl/test/cu121 
 
 # Install two language tokenizers for Translation with TorchText tutorial
 python -m spacy download en_core_web_sm

.jenkins/metadata.json

@@ -29,7 +29,7 @@
     "needs": "linux.16xlarge.nvidia.gpu"
   },
   "intermediate_source/torchvision_tutorial.py": {
-    "needs": "linux.g5.4xlarge.nvidia.gpu", 
+    "needs": "linux.g5.4xlarge.nvidia.gpu",
     "_comment": "does not require a5g but needs to run before gpu_quantization_torchao_tutorial.py."
   },
   "advanced_source/coding_ddpg.py": {
@@ -39,9 +39,18 @@
   "intermediate_source/torch_compile_tutorial.py": {
     "needs": "linux.g5.4xlarge.nvidia.gpu"
   },
+  "intermediate_source/torch_export_tutorial.py": {
+    "needs": "linux.g5.4xlarge.nvidia.gpu"
+  },
   "intermediate_source/scaled_dot_product_attention_tutorial.py": {
     "needs": "linux.g5.4xlarge.nvidia.gpu"
   },
+  "recipes_source/torch_compile_user_defined_triton_kernel_tutorial.py": {
+    "needs": "linux.g5.4xlarge.nvidia.gpu"
+  },
+  "advanced_source/semi_structured_sparse.py": {
+    "needs": "linux.g5.4xlarge.nvidia.gpu"
+  },
   "prototype_source/gpu_quantization_torchao_tutorial.py": {
     "needs": "linux.g5.4xlarge.nvidia.gpu"
   }

.jenkins/validate_tutorials_built.py

@@ -22,12 +22,16 @@
     "beginner_source/former_torchies/tensor_tutorial_old",
     "beginner_source/examples_autograd/polynomial_autograd",
     "beginner_source/examples_autograd/polynomial_custom_function",
+    "beginner_source/torchtext_custom_dataset_tutorial", # not building with 2.3 RC, might be able to turn on with GA
+    "beginner_source/text_sentiment_ngrams_tutorial", # not building with 2.3 RC, might be able to turn on with GA
     "beginner_source/t5_tutorial", # re-enable after this is fixed: https://github.com/pytorch/text/issues/1756
     "intermediate_source/mnist_train_nas",  # used by ax_multiobjective_nas_tutorial.py
+    "intermediate_source/torchvision_tutorial", # disable due to RuntimeError: DataLoader worker (pid(s) 20092) exited unexpectedly
     "intermediate_source/fx_conv_bn_fuser",
     "intermediate_source/_torch_export_nightly_tutorial",  # does not work on release
     "advanced_source/super_resolution_with_onnxruntime",
     "advanced_source/ddp_pipeline",  # requires 4 gpus
+    "advanced_source/usb_semisup_learn", # fails with CUDA OOM error, should try on a different worker
     "prototype_source/fx_graph_mode_ptq_dynamic",
     "prototype_source/vmap_recipe",
     "prototype_source/torchscript_freezing",

_static/img/ExecuTorch-Logo-cropped.svg

@@ -108,7 +108,7 @@
 </noscript>
 
 <script type="text/javascript">
-  var collapsedSections = ['PyTorch Recipes', 'Learning PyTorch', 'Image and Video', 'Audio', 'Text', 'Backends', 'Reinforcement Learning', 'Deploying PyTorch Models in Production', 'Code Transforms with FX', 'Frontend APIs', 'Extending PyTorch', 'Model Optimization', 'Parallel and Distributed Training', 'Mobile'];
+  var collapsedSections = ['PyTorch Recipes', 'Learning PyTorch', 'Image and Video', 'Audio', 'Text', 'Backends', 'Reinforcement Learning', 'Deploying PyTorch Models in Production', 'Profiling PyTorch', 'Code Transforms with FX', 'Frontend APIs', 'Extending PyTorch', 'Model Optimization', 'Parallel and Distributed Training', 'Edge with ExecuTorch', 'Recommendation Systems', 'Multimodality'];
 </script>
 
 <img height="1" width="1" style="border-style:none;" alt="" src="https://www.googleadservices.com/pagead/conversion/795629140/?label=txkmCPmdtosBENSssfsC&amp;guid=ON&amp;script=0"/>

_static/img/distributed/fsdp_tp.png

@@ -226,7 +226,7 @@ The primary goal of integrating new devices through ``PrivateUse1`` is to meet t
 and the next thing to do is to improve usability, which mainly involves the following aspects.
 
 1. Register new backend module to Pytorch.
-2. Generate methods and properties related to the new backend.
+2. Rename PrivateUse1 to a custom name for the new backend.
 3. Generate methods and properties related to the new backend.
 
 Register new backend module to Pytorch