Remove contiguous() in split_embedding_weights_with_scale_bias #1808
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: Calling `tensor.contiguous()` in case of non-contiguous tensor creates a new tensor. Changing it will not change the original `tensor`. To use results of `split_embedding_weights_with_scale_bias(split_scale_bias_mode=2)` as a tensor in state_dict - we should be able via that tensor to change the original tbe weight. For that we need to remove copy via contiguous(). Differential Revision: D46483112 Privacy Context Container: L1138451 fbshipit-source-id: dffaad9c7e92aaaf7761958c0d190a7ed21ee00e
✅ Deploy Preview for pytorch-fbgemm-docs canceled.
|
|
This pull request was exported from Phabricator. Differential Revision: D46483112 |
|
This pull request has been merged in 10e93a0. |
coconutruben
added a commit
to coconutruben/FBGEMM-1
that referenced
this pull request
Jan 23, 2025
Summary: # Why get improvements for MoE kernel # What cherry-pick the relevant components of https://github.com/ROCm/composable_kernel/pull/1808/files# Reviewed By: jianyuh Differential Revision: D68534277
coconutruben
added a commit
to coconutruben/FBGEMM-1
that referenced
this pull request
Jan 25, 2025
Summary: X-link: facebookresearch/FBGEMM#687 # Why get improvements for MoE kernel # What cherry-pick the relevant components of https://github.com/ROCm/composable_kernel/pull/1808/files# Reviewed By: jianyuh Differential Revision: D68534277
coconutruben
added a commit
to coconutruben/FBGEMM-1
that referenced
this pull request
Jan 27, 2025
Summary: X-link: facebookresearch/FBGEMM#687 # Why get improvements for MoE kernel # What cherry-pick the relevant components of https://github.com/ROCm/composable_kernel/pull/1808/files# Reviewed By: jianyuh Differential Revision: D68534277
facebook-github-bot
pushed a commit
that referenced
this pull request
Jan 27, 2025
Summary: Pull Request resolved: #3609 X-link: facebookresearch/FBGEMM#687 # Why get improvements for MoE kernel # What cherry-pick the relevant components of https://github.com/ROCm/composable_kernel/pull/1808/files# Reviewed By: jianyuh Differential Revision: D68534277 fbshipit-source-id: 47a7ce8a239eaa0cd4dc9e713e1603ee3c42825e
coconutruben
added a commit
to coconutruben/FBGEMM-1
that referenced
this pull request
Jan 28, 2025
Summary: # Why - some kernels were missed in the previous cherry-pick - the interface has changed slightly and extracting the `intermediate_size` is different # What - get remaining files from ROCm/composable_kernel@1ff50e7 - adjust interface - explicitly pass silu as the activation function Differential Revision: D68792360
q10
pushed a commit
to q10/FBGEMM
that referenced
this pull request
Apr 10, 2025
Summary: X-link: pytorch#3609 Pull Request resolved: facebookresearch/FBGEMM#687 # Why get improvements for MoE kernel # What cherry-pick the relevant components of https://github.com/ROCm/composable_kernel/pull/1808/files# Reviewed By: jianyuh Differential Revision: D68534277 fbshipit-source-id: 47a7ce8a239eaa0cd4dc9e713e1603ee3c42825e
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:
Calling
tensor.contiguous()in case of non-contiguous tensor creates a new tensor.Changing it will not change the original
tensor.To use results of
split_embedding_weights_with_scale_bias(split_scale_bias_mode=2)as a tensor in state_dict - we should be able via that tensor to change the original tbe weight.For that we need to remove copy via contiguous().
Differential Revision:
D46483112
Privacy Context Container: L1138451