Change set_learning_rate_tensor #3945
Closed
Conversation
This file contains hidden or 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
Summary: The change in D71010630 breaks `aps_models/examples/dlrm/tutorials:test_kernel_apf_dlrm_with_basic_training_demo`, which potentially breaks `apf_dlrm` bento kernel. ``` RuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation: [torch.FloatTensor []] is at version 2; expected version 1 instead. Hint: enable anomaly detection to find the operation that failed to compute its gradient, with torch.autograd.set_detect_anomaly(True). ``` See [Full error log](https://www.internalfb.com/intern/everpaste/?handle=GHmNRBnyyiCXTVkCAP28ClWARWwPbswMAAAz) TBE has a method to set learning rate, i.e., `set_learning_rate(lr)` where `lr` is the learning rate value to be set. D71010630 removes `optimizer_args.learning_rate` (float) and introduces `self.learning_rate_tensor` (tensor). Hence setting learning rate value means changing the value of the `leanring_rate_tensor`. We changed this by using `tensor._fill(lr)`. However, this seems to break bento kernel which is built from APF code, which causes issues when an in-place operation occurs i.e., `tensor._fill(lr)`. The workaround is to create a new tensor to avoid the in-place operation. The change passes the test https://www.internalfb.com/intern/testinfra/testconsole/testrun/3659174972188704/ Reviewed By: sryap, nautsimon Differential Revision: D72617537
|
This pull request was exported from Phabricator. Differential Revision: D72617537 |
✅ Deploy Preview for pytorch-fbgemm-docs ready!
To edit notification comments on pull requests, go to your Netlify site configuration. |
|
This pull request has been merged in 7bb8db2. |
q10
pushed a commit
to q10/FBGEMM
that referenced
this pull request
Apr 10, 2025
Summary: Pull Request resolved: facebookresearch/FBGEMM#1030 X-link: pytorch#3945 The change in D71010630 breaks `aps_models/examples/dlrm/tutorials:test_kernel_apf_dlrm_with_basic_training_demo`, which potentially breaks `apf_dlrm` bento kernel. ``` RuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation: [torch.FloatTensor []] is at version 2; expected version 1 instead. Hint: enable anomaly detection to find the operation that failed to compute its gradient, with torch.autograd.set_detect_anomaly(True). ``` See [Full error log](https://www.internalfb.com/intern/everpaste/?handle=GHmNRBnyyiCXTVkCAP28ClWARWwPbswMAAAz) TBE has a method to set learning rate, i.e., `set_learning_rate(lr)` where `lr` is the learning rate value to be set. D71010630 removes `optimizer_args.learning_rate` (float) and introduces `self.learning_rate_tensor` (tensor). Hence setting learning rate value means changing the value of the `leanring_rate_tensor`. We changed this by using `tensor._fill(lr)`. However, this seems to break bento kernel which is built from APF code, which causes issues when an in-place operation occurs i.e., `tensor._fill(lr)`. The workaround is to create a new tensor to avoid the in-place operation. The change passes the test https://www.internalfb.com/intern/testinfra/testconsole/testrun/3659174972188704/ Reviewed By: sryap, nautsimon Differential Revision: D72617537 fbshipit-source-id: d3b84e872e6b68d7c2a7da5b1380e3d355393c94
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.
Summary:
The change in D71010630 breaks
aps_models/examples/dlrm/tutorials:test_kernel_apf_dlrm_with_basic_training_demo, which potentially breaksapf_dlrmbento kernel.See Full error log
TBE has a method to set learning rate, i.e.,
set_learning_rate(lr)wherelris the learning rate value to be set.D71010630 removes
optimizer_args.learning_rate(float) and introducesself.learning_rate_tensor(tensor). Hence setting learning rate value means changing the value of theleanring_rate_tensor. We changed this by usingtensor._fill(lr).However, this seems to break bento kernel which is built from APF code, which causes issues when an in-place operation occurs i.e.,
tensor._fill(lr).The workaround is to create a new tensor to avoid the in-place operation. The change passes the test
https://www.internalfb.com/intern/testinfra/testconsole/testrun/3659174972188704/
Reviewed By: sryap, nautsimon
Differential Revision: D72617537