[pull] main from pytorch:main #238
Open
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
![pull[bot]](https://avatars.githubusercontent.com/in/12910?s=60&v=4){kind=small}
* Fix up custom op tutorials --------- Co-authored-by: Svetlana Karslioglu <[email protected]>
* Added the first version of LR sched tutorial --------- Co-authored-by: Svetlana Karslioglu <[email protected]>
* Update formatting in the autograd_tutorial.py --------- Co-authored-by: Svetlana Karslioglu <[email protected]>
Explicitly mention what the import for `Random()` is in comment
Co-authored-by: Svetlana Karslioglu <[email protected]>
Batch size must be an int, not a float. This change fixes it, basically doing the same as in https://github.com/seba-1511/dist_tuto.pth/blob/a552567061a9985cdcfe72ecb9b47e4630d6a7fe/train_dist.py#L85. Co-authored-by: Svetlana Karslioglu <[email protected]>
Fix indent show copy of code doesn't include tick marks Co-authored-by: Svetlana Karslioglu <[email protected]>
* DEP: deprecate transformer tutorial * Update beginner_source/transformer_tutorial.rst * Update beginner_source/transformer_tutorial.rst --------- Co-authored-by: Svetlana Karslioglu <[email protected]>
* Remove torchvision_tutorial from validate_tutorials_built.py * Fix multiprocessing in torchvision_tutorial.py --------- Co-authored-by: Svetlana Karslioglu <[email protected]>
* [Docs] Update performance tuning guide Added cuda graph explaination Added core pinning section Added tensor core usage section --------- Co-authored-by: Svetlana Karslioglu <[email protected]>
Co-authored-by: Svetlana Karslioglu <[email protected]>
Many overdue updates * updating the overview to include TP/PP and DTensor/Devicemesh * removing RPC, DataParallel and Elastic as they are no longer supported
Didn't realize that this was a convention, so restoring it
…ders (#2922) * add redirect to data tutorial --------- Co-authored-by: Svetlana Karslioglu <[email protected]>
* Add new Python and C++/CUDA Custom Op tutorials I want to land this before PyTorch 2.4 (so we can link to these in PyTorch's nightly documentation) and then have a follow-up PR for 2.4 that actually runs the scripts (so that they can generate outputs). pytorch/pytorch#127443 to remind myself of the above. NB: These two tutorials replace all of the existing custom ops (and cpp extensions) tutorials: - advanced/cpp_extension - advanced/torch_script_custom_ops - advanced/torch_script_custom_classes - advanced/dispatcher In a follow-up PR we will add warnings to all of those tutorials stating that they are deprecated but we will preserve the text just in case people still need them (e.g. if they are not using PyTorch 2.4). Test Plan: - I tested these locally. * Fix cpp custom ops tutorial per review * update * update index.rst * Update * update * update
Co-authored-by: Svetlana Karslioglu <[email protected]>
* added timing comparison --------- Co-authored-by: Svetlana Karslioglu <[email protected]>
Co-authored-by: Svetlana Karslioglu <[email protected]>
* Recommend python agnosticism in cpp custom op tutorial * forgot to delete a line * Fixed tutorial to be clearer and to recommend other path * Switch to commits instead of master * Formatting code blocks * polish + i missed some code blocks earlier * Adjust advice based on Sam Gross's knowlede
.rst, .md, and .py file changes are included for testing purposes only and will not be merged. Mering despite of the lint errors, could be fixed in followup PRs I guess
* Update what's new in tutorials
* Move back to use stable binaries 2.6
* change torch.cuda.amp.GradScaler to torch.GradScaler("cuda")
* changing torch.GradScaler to torch.amp.GradScaler
---------
Co-authored-by: Svetlana Karslioglu <[email protected]>
Add: - DeviceMesh - intra - composability - HSDP
* fixing typo "sequance->sequence" --------- Co-authored-by: Svetlana Karslioglu <[email protected]>
* Add a new tutorial for compile time caching --------- Co-authored-by: Svetlana Karslioglu <[email protected]>
--------- Co-authored-by: Svetlana Karslioglu <[email protected]>
* add and fix custom cards for ONNX
Co-authored-by: Svetlana Karslioglu <[email protected]>
RandomSizedCrop depreciated in favor of RandomResizedCrop in 0.9 https://pytorch.org/vision/0.9/transforms.html Co-authored-by: Svetlana Karslioglu <[email protected]>
* Create MonthlyLinkCheck.yml * Update MonthlyLinkCheck.yml * Update MonthlyLinkCheck.yml * Update MonthlyLinkCheck.yml
Add Docs Survey to Tutorials Site. Will be disabled on April 14, 2025
Improves char-rnn tutorial code quality ## Description This PR adds proper handling for Out-Of-Vocabulary (OOV) characters in the character-level RNN tutorial. Problem: - The current implementation doesn't properly handle characters not in the allowed set - Using string.find() returns -1 for unknown characters, causing them to be treated as apostrophes (the last character in the allowed_characters string) - This creates ambiguity between actual apostrophes in names (like O'Brien) and unknown characters Solution: - Added an underscore character "_" as a dedicated OOV token - Modified letterToIndex() to explicitly handle unknown characters - Added comments explaining the purpose of OOV handling - Updated the comment about input nodes (57 → 58) to reflect the added character This change follows best practices for NLP systems by explicitly handling unknown characters, improving both the model's accuracy and the tutorial's educational value.
The tags in tj-actions/changed-files action are compromised and are leaking GitHub secrets in repos using the compromised repo. This pins the action to a known good hash. https://www.stepsecurity.io/blog/harden-runner-detection-tj-actions-changed-files-action-is-compromised Signed-off-by: Thanh Ha <[email protected]>
Co-authored-by: Vincent Moens <[email protected]>
Attempt to use a better regex.
We tightned invariants in PyTorch 2.7 that exposed a bug with the python custom ops tutorial. This PR fixes said bug. Test Plan: - tested locally
* Update C++ Extension tutorial with note about AMD GPU. --------- Co-authored-by: Svetlana Karslioglu <[email protected]>
* Bump torchrl and torchdict to 0.7.2 * Add devices * fix semi_structured_sparse.py with default device * Disable semi sparse tutorial --------- Co-authored-by: Vincent Moens <[email protected]>
Fix typos --------- Co-authored-by: Svetlana Karslioglu <[email protected]>
* add system requirements to compile tutorial * Apply suggestions from code review Co-authored-by: Svetlana Karslioglu <[email protected]> --------- Co-authored-by: Svetlana Karslioglu <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
See Commits and Changes for more details.
Created by
pull[bot]
Can you help keep this open source service alive? 💖 Please sponsor : )