New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[Gradient Compression] Add an index field to GradBucket for PowerSGD #48757
Commits on Dec 3, 2020
-
[Gradient Compression] Add an index field to GradBucket for PowerSGD
Add an index field to GradBucekt, so error_dict is keyed by this index instead of the hashcode of input tensor. Howevever, sometimes the buckets can be rebuilt in the forward pass. In this case, the shape of the bucket with the same index will not be consistent with the one in the previous iteration, and hence the error tensor will be re--initialized as a zero tensor of the new shape. Original PR issue: Investigate Applying PowerSGD to Communication Hook for Gradient Compression #47202 Differential Revision: [D25288496](https://our.internmc.facebook.com/intern/diff/D25288496/) [ghstack-poisoned]
wayi committedDec 3, 2020 Configuration menu - View commit details
-
Copy full SHA for f5cb18b - Browse repository at this point
Copy the full SHA f5cb18bView commit details -
Update on "[Gradient Compression] Add an index field to GradBucket fo…
…r PowerSGD" Add an index field to GradBucekt, so error_dict is keyed by this index instead of the hashcode of input tensor. Howevever, sometimes the buckets can be rebuilt in the forward pass. In this case, the shape of the bucket with the same index will not be consistent with the one in the previous iteration, and hence the error tensor will be re--initialized as a zero tensor of the new shape. Original PR issue: Investigate Applying PowerSGD to Communication Hook for Gradient Compression #47202 Differential Revision: [D25288496](https://our.internmc.facebook.com/intern/diff/D25288496/) [ghstack-poisoned]
wayi committedDec 3, 2020 Configuration menu - View commit details
-
Copy full SHA for d1ae9af - Browse repository at this point
Copy the full SHA d1ae9afView commit details
Commits on Dec 4, 2020
-
Update on "[Gradient Compression] Add an index field to GradBucket fo…
…r PowerSGD" Add an index field to GradBucekt, so error_dict is keyed by this index instead of the hashcode of input tensor. Howevever, sometimes the buckets can be rebuilt in the forward pass. In this case, the shape of the bucket with the same index will not be consistent with the one in the previous iteration, and hence the error tensor will be re--initialized as a zero tensor of the new shape. Original PR issue: Investigate Applying PowerSGD to Communication Hook for Gradient Compression #47202 Differential Revision: [D25288496](https://our.internmc.facebook.com/intern/diff/D25288496/) [ghstack-poisoned]
wayi committedDec 4, 2020 Configuration menu - View commit details
-
Copy full SHA for ee4f2a8 - Browse repository at this point
Copy the full SHA ee4f2a8View commit details -
Update on "[Gradient Compression] Add an index field to GradBucket fo…
…r PowerSGD" Add an index field to GradBucekt, so error_dict is keyed by this index instead of the hashcode of input tensor. Howevever, sometimes the buckets can be rebuilt in the forward pass. In this case, the shape of the bucket with the same index will not be consistent with the one in the previous iteration, and hence the error tensor will be re--initialized as a zero tensor of the new shape. Original PR issue: Investigate Applying PowerSGD to Communication Hook for Gradient Compression #47202 Differential Revision: [D25288496](https://our.internmc.facebook.com/intern/diff/D25288496/) [ghstack-poisoned]
wayi committedDec 4, 2020 Configuration menu - View commit details
-
Copy full SHA for de1a339 - Browse repository at this point
Copy the full SHA de1a339View commit details
Commits on Dec 5, 2020
-
Update on "[Gradient Compression] Add an index field to GradBucket fo…
…r PowerSGD" Add an index field to GradBucekt, so error_dict is keyed by this index instead of the hashcode of input tensor. Howevever, sometimes the buckets can be rebuilt in the forward pass. In this case, the shape of the bucket with the same index will not be consistent with the one in the previous iteration, and hence the error tensor will be re--initialized as a zero tensor of the new shape. Original PR issue: Investigate Applying PowerSGD to Communication Hook for Gradient Compression #47202 Differential Revision: [D25288496](https://our.internmc.facebook.com/intern/diff/D25288496/) [ghstack-poisoned]
wayi committedDec 5, 2020 Configuration menu - View commit details
-
Copy full SHA for 5cafdf8 - Browse repository at this point
Copy the full SHA 5cafdf8View commit details -
Update on "[Gradient Compression] Add an index field to GradBucket fo…
…r PowerSGD" Add an index field to GradBucekt, so error_dict is keyed by this index instead of the hashcode of input tensor. Howevever, sometimes the buckets can be rebuilt in the forward pass. In this case, the shape of the bucket with the same index will not be consistent with the one in the previous iteration, and hence the error tensor will be re--initialized as a zero tensor of the new shape. Original PR issue: Investigate Applying PowerSGD to Communication Hook for Gradient Compression #47202 Differential Revision: [D25288496](https://our.internmc.facebook.com/intern/diff/D25288496/) [ghstack-poisoned]
wayi committedDec 5, 2020 Configuration menu - View commit details
-
Copy full SHA for c05592d - Browse repository at this point
Copy the full SHA c05592dView commit details