Propagate device type for heterogenous sharding of table across different device types #2606
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
facebook-github-bot
added
the
CLA Signed
|
This pull request was exported from Phabricator. Differential Revision: D66682124 |
faran928
force-pushed
the
export-D66682124
branch
from
dcac3ec to
0284270
Compare
|
This pull request was exported from Phabricator. Differential Revision: D66682124 |
faran928
force-pushed
the
export-D66682124
branch
from
0284270 to
19153a9
Compare
faran928
added a commit
to faran928/torchrec
that referenced
this pull request
Dec 5, 2024
…rent device types (pytorch#2606) Summary: For row wise heterogenous sharding of tables acorss cuda and cpu, we propagate the correct device type within each look up module based on which shard of each table is being looked up / fetched within that module. We also move some of the wrapper functions that can enable us to pass batch info information correctly across different modules during model split. The changes should be backward compatible and not impact existing behavior Differential Revision: D66682124
|
This pull request was exported from Phabricator. Differential Revision: D66682124 |
faran928
force-pushed
the
export-D66682124
branch
from
19153a9 to
c383a6d
Compare
faran928
added a commit
to faran928/torchrec
that referenced
this pull request
Dec 18, 2024
…rent device types (pytorch#2606) Summary: For row wise heterogenous sharding of tables acorss cuda and cpu, we propagate the correct device type within each look up module based on which shard of each table is being looked up / fetched within that module. We also move some of the wrapper functions that can enable us to pass batch info information correctly across different modules during model split. The changes should be backward compatible and not impact existing behavior Differential Revision: D66682124
|
This pull request was exported from Phabricator. Differential Revision: D66682124 |
faran928
added a commit
to faran928/torchrec
that referenced
this pull request
Dec 18, 2024
…rent device types (pytorch#2606) Summary: For row wise heterogenous sharding of tables acorss cuda and cpu, we propagate the correct device type within each look up module based on which shard of each table is being looked up / fetched within that module. We also move some of the wrapper functions that can enable us to pass batch info information correctly across different modules during model split. The changes should be backward compatible and not impact existing behavior Differential Revision: D66682124
faran928
force-pushed
the
export-D66682124
branch
from
c383a6d to
c006ee4
Compare
|
This pull request was exported from Phabricator. Differential Revision: D66682124 |
faran928
force-pushed
the
export-D66682124
branch
from
c006ee4 to
8f7b9ae
Compare
faran928
added a commit
to faran928/torchrec
that referenced
this pull request
Dec 20, 2024
…rent device types (pytorch#2606) Summary: For row wise heterogenous sharding of tables acorss cuda and cpu, we propagate the correct device type within each look up module based on which shard of each table is being looked up / fetched within that module. We also move some of the wrapper functions that can enable us to pass batch info information correctly across different modules during model split. The changes should be backward compatible and not impact existing behavior Differential Revision: D66682124
|
This pull request was exported from Phabricator. Differential Revision: D66682124 |
faran928
added a commit
to faran928/torchrec
that referenced
this pull request
Dec 20, 2024
…rent device types (pytorch#2606) Summary: For row wise heterogenous sharding of tables acorss cuda and cpu, we propagate the correct device type within each look up module based on which shard of each table is being looked up / fetched within that module. We also move some of the wrapper functions that can enable us to pass batch info information correctly across different modules during model split. The changes should be backward compatible and not impact existing behavior Reviewed By: jiayisuse Differential Revision: D66682124
faran928
force-pushed
the
export-D66682124
branch
from
8f7b9ae to
fdcc59e
Compare
|
This pull request was exported from Phabricator. Differential Revision: D66682124 |
…rent device types (pytorch#2606) Summary: For row wise heterogenous sharding of tables acorss cuda and cpu, we propagate the correct device type within each look up module based on which shard of each table is being looked up / fetched within that module. We also move some of the wrapper functions that can enable us to pass batch info information correctly across different modules during model split. The changes should be backward compatible and not impact existing behavior Reviewed By: jiayisuse Differential Revision: D66682124
faran928
force-pushed
the
export-D66682124
branch
from
fdcc59e to
f6b54cd
Compare
|
This pull request was exported from Phabricator. Differential Revision: D66682124 |
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:
For row wise heterogenous sharding of tables acorss cuda and cpu, we propagate the correct device type within each look up module based on which shard of each table is being looked up / fetched within that module.
The changes should be backward compatible and not impact existing behavior
Differential Revision: D66682124