fix embedding_backward_dense decomp with broadcasting #95499
Closed
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
[ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/95499
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 4919634: This comment was automatically generated by Dr. CI and updates every 15 minutes. |
bdhirsh
added a commit
that referenced
this pull request
Feb 24, 2023
ghstack-source-id: 411189909c064677d8d2783af7b19bbb80577ee0 Pull Request resolved: #95499
github-actions
bot
requested review from
albanD,
antoniojkim,
Chillee,
ezyang,
jbschlosser,
miladm,
SherlockNoMad,
voznesenskym and
wconstab
ezyang
approved these changes
Feb 24, 2023
Fixes #95182 cc ngimel for another decomp fix. For this one, I tried auditing the CPU and CUDA kernels for `embedding_backward_dense` and just could not figure out where the `unsqueeze(1)` was supposed to be coming from. In the failing example, our tensor shapes are `(2, 4, 3)` and `(2, 4)`, and so I just assumed that the existing decomp had a typo - we should be unsqueezing the last dim, instead of dim index 1. That fixes the repro, and the existing decomp + meta tests appear to be passing. cc soumith voznesenskym yanboliang penguinwu anijain2305 EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng Xia-Weiwen wenzhe-nrv jiayisunx desertfire [ghstack-poisoned]
bdhirsh
added a commit
that referenced
this pull request
Feb 24, 2023
ghstack-source-id: dfea7c497fde1ad0c3030bc7bc6fe6e790de8954 Pull Request resolved: #95499
Fixes #95182 cc soumith voznesenskym yanboliang penguinwu anijain2305 EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng Xia-Weiwen wenzhe-nrv jiayisunx desertfire ngimel for another decomp fix. For this one, I tried auditing the CPU and CUDA kernels for `embedding_backward_dense` and just could not figure out where the `unsqueeze(1)` was supposed to be coming from. In the failing example, our tensor shapes are `(2, 4, 3)` and `(2, 4)`, and so I just assumed that the existing decomp had a typo - we should be unsqueezing the last dim, instead of dim index 1. That fixes the repro, and the existing decomp + meta tests appear to be passing. cc soumith voznesenskym yanboliang penguinwu anijain2305 EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng Xia-Weiwen wenzhe-nrv jiayisunx desertfire [ghstack-poisoned]
cyyever
pushed a commit
to cyyever/pytorch_private
that referenced
this pull request
Mar 1, 2023
Fixes pytorch/pytorch#95182 Pull Request resolved: pytorch/pytorch#95499 Approved by: https://github.com/ezyang, https://github.com/ngimel
cyyever
pushed a commit
to cyyever/pytorch_private
that referenced
this pull request
Mar 2, 2023
Fixes pytorch/pytorch#95182 Pull Request resolved: pytorch/pytorch#95499 Approved by: https://github.com/ezyang, https://github.com/ngimel
cyyever
pushed a commit
to cyyever/pytorch_private
that referenced
this pull request
Mar 5, 2023
Fixes pytorch/pytorch#95182 Pull Request resolved: pytorch/pytorch#95499 Approved by: https://github.com/ezyang, https://github.com/ngimel
cyyever
pushed a commit
to cyyever/pytorch_private
that referenced
this pull request
Mar 5, 2023
Fixes pytorch/pytorch#95182 Pull Request resolved: pytorch/pytorch#95499 Approved by: https://github.com/ezyang, https://github.com/ngimel
cyyever
pushed a commit
to cyyever/pytorch_private
that referenced
this pull request
Mar 27, 2023
Fixes pytorch/pytorch#95182 Pull Request resolved: pytorch/pytorch#95499 Approved by: https://github.com/ezyang, https://github.com/ngimel
jhavukainen
pushed a commit
to kulinseth/pytorch
that referenced
this pull request
Mar 15, 2024
Fixes pytorch#95182 Pull Request resolved: pytorch#95499 Approved by: https://github.com/ezyang, https://github.com/ngimel
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.
Fixes #95182
cc @soumith @voznesenskym @yanboliang @penguinwu @anijain2305 @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @Xia-Weiwen @wenzhe-nrv @jiayisunx @desertfire @ngimel for another decomp fix. For this one, I tried auditing the CPU and CUDA kernels for
embedding_backward_denseand just could not figure out where theunsqueeze(1)was supposed to be coming from. In the failing example, our tensor shapes are(2, 4, 3)and(2, 4), and so I just assumed that the existing decomp had a typo - we should be unsqueezing the last dim, instead of dim index 1. That fixes the repro, and the existing decomp + meta tests appear to be passing.Stack from ghstack (oldest at bottom):
cc @soumith @voznesenskym @yanboliang @penguinwu @anijain2305 @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @Xia-Weiwen @wenzhe-nrv @jiayisunx @desertfire