add DTensor to optimizer state dict #2585
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
facebook-github-bot
added
the
CLA Signed
|
This pull request was exported from Phabricator. Differential Revision: D65555455 |
iamzainhuda
force-pushed
the
export-D65555455
branch
from
9de67ab to
9aab41c
Compare
iamzainhuda
added a commit
to iamzainhuda/torchrec
that referenced
this pull request
Dec 9, 2024
Summary: To support 2D parallelism checkpointing, we introduce DTensor to the optimizer state dict. It is enabled through fused_params["output_dtensor"] = True, meaning when table shards are outputted in DTensor so are optimizer shards. This diff allows us to leverage N-dimensional device meshes with support for abritrary replication/sharding groups - making checkpointing easy as DCP/Modelstore support replicated/sharded placements on a device mesh (something that is unsupported in ShardedTensor) Differential Revision: D65555455
|
This pull request was exported from Phabricator. Differential Revision: D65555455 |
Summary: To support 2D parallelism checkpointing, we introduce DTensor to the optimizer state dict. It is enabled through fused_params["output_dtensor"] = True, meaning when table shards are outputted in DTensor so are optimizer shards. This diff allows us to leverage N-dimensional device meshes with support for abritrary replication/sharding groups - making checkpointing easy as DCP/Modelstore support replicated/sharded placements on a device mesh (something that is unsupported in ShardedTensor) Differential Revision: D65555455
iamzainhuda
force-pushed
the
export-D65555455
branch
from
9aab41c to
6cb7df9
Compare
|
This pull request was exported from Phabricator. Differential Revision: D65555455 |
iamzainhuda
added a commit
to iamzainhuda/torchrec
that referenced
this pull request
Dec 10, 2024
Summary: To support 2D parallelism checkpointing, we introduce DTensor to the optimizer state dict. It is enabled through fused_params["output_dtensor"] = True, meaning when table shards are outputted in DTensor so are optimizer shards. This diff allows us to leverage N-dimensional device meshes with support for abritrary replication/sharding groups - making checkpointing easy as DCP/Modelstore support replicated/sharded placements on a device mesh (something that is unsupported in ShardedTensor) Differential Revision: D65555455
iamzainhuda
added a commit
to iamzainhuda/torchrec
that referenced
this pull request
Dec 10, 2024
Summary: To support 2D parallelism checkpointing, we introduce DTensor to the optimizer state dict. It is enabled through fused_params["output_dtensor"] = True, meaning when table shards are outputted in DTensor so are optimizer shards. This diff allows us to leverage N-dimensional device meshes with support for abritrary replication/sharding groups - making checkpointing easy as DCP/Modelstore support replicated/sharded placements on a device mesh (something that is unsupported in ShardedTensor) Differential Revision: D65555455
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:
To support 2D parallelism checkpointing, we introduce DTensor to the optimizer state dict.
This diff allows us to leverage N-dimensional device meshes with support for abritrary replication/sharding groups - making checkpointing easy as DCP support replicated/sharded placements on a device mesh (something that is unsupported in ShardedTensor)
Differential Revision: D65555455