Add keep_orig_idx_per_feature parameter to block_bucketize_sparse_features kernel #4027
+320
−38
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
|
This pull request was exported from Phabricator. Differential Revision: D73606958 |
✅ Deploy Preview for pytorch-fbgemm-docs ready!
To edit notification comments on pull requests, go to your Netlify site configuration. |
emlin
added a commit
to emlin/FBGEMM
that referenced
this pull request
Apr 26, 2025
…tures kernel (pytorch#4027) Summary: X-link: facebookresearch/FBGEMM#1112 **Context** Enhance block_bucketize_sparse_features and block_bucketize_sparse_features_inference kernels to support mixed-format embedding tables. Previously, the keep_orig_idx parameter was a boolean flag applied uniformly across all features, determining whether to retain the original index. With the introduction of [the Flexible Collision-Free Embedding Table](https://github.com/pytorch/torchrec/blob/main/rfc/RFC-0002-Flexible-Collision-Free-Embedding-Table.md), one embedding collection may include both collision-free and collision tables. This update allows the kernel to handle mixed formats by supporting feature-wise control over index retention. For collision-free tables, a large table size of 2^50 is set, maintaining parameters as id-value pairs and preserving the original global id. This change facilitates the use of mixed-style embedding tables effectively. Spec: - keep_orig_idx_per_feature is an optional parameter with per feature settings. - If the keep_orig_idx_per_feature is not None, the value will override global flag keep_orig_idx, no matter it's true for false. - If keep_orig_idx_per_feature is None, fallback to keep_orig_idx control. Note: Adding additional parameter keep_orig_idx_per_feature, instead of change keep_orig_idx directly, is to avoid backward compatibility issue. Differential Revision: D73606958
emlin
added a commit
to emlin/FBGEMM
that referenced
this pull request
Apr 26, 2025
…tures kernel (pytorch#4027) Summary: X-link: facebookresearch/FBGEMM#1112 **Context** Enhance block_bucketize_sparse_features and block_bucketize_sparse_features_inference kernels to support mixed-format embedding tables. Previously, the keep_orig_idx parameter was a boolean flag applied uniformly across all features, determining whether to retain the original index. With the introduction of [the Flexible Collision-Free Embedding Table](https://github.com/pytorch/torchrec/blob/main/rfc/RFC-0002-Flexible-Collision-Free-Embedding-Table.md), one embedding collection may include both collision-free and collision tables. This update allows the kernel to handle mixed formats by supporting feature-wise control over index retention. For collision-free tables, a large table size of 2^50 is set, maintaining parameters as id-value pairs and preserving the original global id. This change facilitates the use of mixed-style embedding tables effectively. Spec: - keep_orig_idx_per_feature is an optional parameter with per feature settings. - If the keep_orig_idx_per_feature is not None, the value will override global flag keep_orig_idx, no matter it's true for false. - If keep_orig_idx_per_feature is None, fallback to keep_orig_idx control. Note: Adding additional parameter keep_orig_idx_per_feature, instead of change keep_orig_idx directly, is to avoid backward compatibility issue. Differential Revision: D73606958
|
This pull request was exported from Phabricator. Differential Revision: D73606958 |
emlin
added a commit
to emlin/FBGEMM
that referenced
this pull request
Apr 26, 2025
…tures kernel (pytorch#4027) Summary: Pull Request resolved: pytorch#4027 X-link: facebookresearch/FBGEMM#1112 **Context** Enhance block_bucketize_sparse_features and block_bucketize_sparse_features_inference kernels to support mixed-format embedding tables. Previously, the keep_orig_idx parameter was a boolean flag applied uniformly across all features, determining whether to retain the original index. With the introduction of [the Flexible Collision-Free Embedding Table](https://github.com/pytorch/torchrec/blob/main/rfc/RFC-0002-Flexible-Collision-Free-Embedding-Table.md), one embedding collection may include both collision-free and collision tables. This update allows the kernel to handle mixed formats by supporting feature-wise control over index retention. For collision-free tables, a large table size of 2^50 is set, maintaining parameters as id-value pairs and preserving the original global id. This change facilitates the use of mixed-style embedding tables effectively. Spec: - keep_orig_idx_per_feature is an optional parameter with per feature settings. - If the keep_orig_idx_per_feature is not None, the value will override global flag keep_orig_idx, no matter it's true for false. - If keep_orig_idx_per_feature is None, fallback to keep_orig_idx control. Note: Adding additional parameter keep_orig_idx_per_feature, instead of change keep_orig_idx directly, is to avoid backward compatibility issue. Differential Revision: D73606958
|
This pull request was exported from Phabricator. Differential Revision: D73606958 |
emlin
added a commit
to emlin/FBGEMM
that referenced
this pull request
Apr 26, 2025
…tures kernel (pytorch#4027) Summary: Pull Request resolved: pytorch#4027 X-link: facebookresearch/FBGEMM#1112 **Context** Enhance block_bucketize_sparse_features and block_bucketize_sparse_features_inference kernels to support mixed-format embedding tables. Previously, the keep_orig_idx parameter was a boolean flag applied uniformly across all features, determining whether to retain the original index. With the introduction of [the Flexible Collision-Free Embedding Table](https://github.com/pytorch/torchrec/blob/main/rfc/RFC-0002-Flexible-Collision-Free-Embedding-Table.md), one embedding collection may include both collision-free and collision tables. This update allows the kernel to handle mixed formats by supporting feature-wise control over index retention. For collision-free tables, a large table size of 2^50 is set, maintaining parameters as id-value pairs and preserving the original global id. This change facilitates the use of mixed-style embedding tables effectively. Spec: - keep_orig_idx_per_feature is an optional parameter with per feature settings. - If the keep_orig_idx_per_feature is not None, the value will override global flag keep_orig_idx, no matter it's true for false. - If keep_orig_idx_per_feature is None, fallback to keep_orig_idx control. Note: Adding additional parameter keep_orig_idx_per_feature, instead of change keep_orig_idx directly, is to avoid backward compatibility issue. Differential Revision: D73606958
emlin
force-pushed
the
export-D73606958
branch
2 times, most recently
from
6c9d912 to
710535a
Compare
|
This pull request was exported from Phabricator. Differential Revision: D73606958 |
emlin
added a commit
to emlin/FBGEMM
that referenced
this pull request
Apr 30, 2025
…tures kernel (pytorch#4027) Summary: X-link: facebookresearch/FBGEMM#1112 **Context** Enhance block_bucketize_sparse_features and block_bucketize_sparse_features_inference kernels to support mixed-format embedding tables. Previously, the keep_orig_idx parameter was a boolean flag applied uniformly across all features, determining whether to retain the original index. With the introduction of [the Flexible Collision-Free Embedding Table](https://github.com/pytorch/torchrec/blob/main/rfc/RFC-0002-Flexible-Collision-Free-Embedding-Table.md), one embedding collection may include both collision-free and collision tables. This update allows the kernel to handle mixed formats by supporting feature-wise control over index retention. For collision-free tables, a large table size of 2^50 is set, maintaining parameters as id-value pairs and preserving the original global id. This change facilitates the use of mixed-style embedding tables effectively. Spec: - keep_orig_idx_per_feature is an optional parameter with per feature settings. - If the keep_orig_idx_per_feature is not None, the value will override global flag keep_orig_idx, no matter it's true for false. - If keep_orig_idx_per_feature is None, fallback to keep_orig_idx control. Note: Adding additional parameter keep_orig_idx_per_feature, instead of change keep_orig_idx directly, is to avoid backward compatibility issue. Differential Revision: D73606958
emlin
added a commit
to emlin/FBGEMM
that referenced
this pull request
Apr 30, 2025
…tures kernel (pytorch#4027) Summary: X-link: facebookresearch/FBGEMM#1112 **Context** Enhance block_bucketize_sparse_features and block_bucketize_sparse_features_inference kernels to support mixed-format embedding tables. Previously, the keep_orig_idx parameter was a boolean flag applied uniformly across all features, determining whether to retain the original index. With the introduction of [the Flexible Collision-Free Embedding Table](https://github.com/pytorch/torchrec/blob/main/rfc/RFC-0002-Flexible-Collision-Free-Embedding-Table.md), one embedding collection may include both collision-free and collision tables. This update allows the kernel to handle mixed formats by supporting feature-wise control over index retention. For collision-free tables, a large table size of 2^50 is set, maintaining parameters as id-value pairs and preserving the original global id. This change facilitates the use of mixed-style embedding tables effectively. Spec: - keep_orig_idx_per_feature is an optional parameter with per feature settings. - If the keep_orig_idx_per_feature is not None, the value will override global flag keep_orig_idx, no matter it's true for false. - If keep_orig_idx_per_feature is None, fallback to keep_orig_idx control. Note: Adding additional parameter keep_orig_idx_per_feature, instead of change keep_orig_idx directly, is to avoid backward compatibility issue. Differential Revision: D73606958
|
This pull request was exported from Phabricator. Differential Revision: D73606958 |
emlin
added a commit
to emlin/FBGEMM
that referenced
this pull request
May 3, 2025
…tures kernel (pytorch#4027) Summary: X-link: facebookresearch/FBGEMM#1112 **Context** Enhance block_bucketize_sparse_features and block_bucketize_sparse_features_inference kernels to support mixed-format embedding tables. Previously, the keep_orig_idx parameter was a boolean flag applied uniformly across all features, determining whether to retain the original index. With the introduction of [the Flexible Collision-Free Embedding Table](https://github.com/pytorch/torchrec/blob/main/rfc/RFC-0002-Flexible-Collision-Free-Embedding-Table.md), one embedding collection may include both collision-free and collision tables. This update allows the kernel to handle mixed formats by supporting feature-wise control over index retention. For collision-free tables, a large table size of 2^50 is set, maintaining parameters as id-value pairs and preserving the original global id. This change facilitates the use of mixed-style embedding tables effectively. Spec: - keep_orig_idx_per_feature is an optional parameter with per feature settings. - If the keep_orig_idx_per_feature is not None, the value will override global flag keep_orig_idx, no matter it's true for false. - If keep_orig_idx_per_feature is None, fallback to keep_orig_idx control. Note: Adding additional parameter keep_orig_idx_per_feature, instead of change keep_orig_idx directly, is to avoid backward compatibility issue. Differential Revision: D73606958
emlin
added a commit
to emlin/FBGEMM
that referenced
this pull request
May 3, 2025
…tures kernel (pytorch#4027) Summary: X-link: facebookresearch/FBGEMM#1112 **Context** Enhance block_bucketize_sparse_features and block_bucketize_sparse_features_inference kernels to support mixed-format embedding tables. Previously, the keep_orig_idx parameter was a boolean flag applied uniformly across all features, determining whether to retain the original index. With the introduction of [the Flexible Collision-Free Embedding Table](https://github.com/pytorch/torchrec/blob/main/rfc/RFC-0002-Flexible-Collision-Free-Embedding-Table.md), one embedding collection may include both collision-free and collision tables. This update allows the kernel to handle mixed formats by supporting feature-wise control over index retention. For collision-free tables, a large table size of 2^50 is set, maintaining parameters as id-value pairs and preserving the original global id. This change facilitates the use of mixed-style embedding tables effectively. Spec: - keep_orig_idx_per_feature is an optional parameter with per feature settings. - If the keep_orig_idx_per_feature is not None, the value will override global flag keep_orig_idx, no matter it's true for false. - If keep_orig_idx_per_feature is None, fallback to keep_orig_idx control. Note: Adding additional parameter keep_orig_idx_per_feature, instead of change keep_orig_idx directly, is to avoid backward compatibility issue. Differential Revision: D73606958
|
This pull request was exported from Phabricator. Differential Revision: D73606958 |
emlin
added a commit
to emlin/FBGEMM
that referenced
this pull request
May 3, 2025
…tures kernel (pytorch#4027) Summary: Pull Request resolved: pytorch#4027 X-link: facebookresearch/FBGEMM#1112 **Context** Enhance block_bucketize_sparse_features and block_bucketize_sparse_features_inference kernels to support mixed-format embedding tables. Previously, the keep_orig_idx parameter was a boolean flag applied uniformly across all features, determining whether to retain the original index. With the introduction of [the Flexible Collision-Free Embedding Table](https://github.com/pytorch/torchrec/blob/main/rfc/RFC-0002-Flexible-Collision-Free-Embedding-Table.md), one embedding collection may include both collision-free and collision tables. This update allows the kernel to handle mixed formats by supporting feature-wise control over index retention. For collision-free tables, a large table size of 2^50 is set, maintaining parameters as id-value pairs and preserving the original global id. This change facilitates the use of mixed-style embedding tables effectively. Spec: - keep_orig_idx_per_feature is an optional parameter with per feature settings. - If the keep_orig_idx_per_feature is not None, the value will override global flag keep_orig_idx, no matter it's true for false. - If keep_orig_idx_per_feature is None, fallback to keep_orig_idx control. Note: Adding additional parameter keep_orig_idx_per_feature, instead of change keep_orig_idx directly, is to avoid backward compatibility issue. Differential Revision: D73606958
|
This pull request was exported from Phabricator. Differential Revision: D73606958 |
emlin
added a commit
to emlin/FBGEMM
that referenced
this pull request
May 3, 2025
…tures kernel (pytorch#4027) Summary: Pull Request resolved: pytorch#4027 X-link: facebookresearch/FBGEMM#1112 **Context** Enhance block_bucketize_sparse_features and block_bucketize_sparse_features_inference kernels to support mixed-format embedding tables. Previously, the keep_orig_idx parameter was a boolean flag applied uniformly across all features, determining whether to retain the original index. With the introduction of [the Flexible Collision-Free Embedding Table](https://github.com/pytorch/torchrec/blob/main/rfc/RFC-0002-Flexible-Collision-Free-Embedding-Table.md), one embedding collection may include both collision-free and collision tables. This update allows the kernel to handle mixed formats by supporting feature-wise control over index retention. For collision-free tables, a large table size of 2^50 is set, maintaining parameters as id-value pairs and preserving the original global id. This change facilitates the use of mixed-style embedding tables effectively. Spec: - keep_orig_idx_per_feature is an optional parameter with per feature settings. - If the keep_orig_idx_per_feature is not None, the value will override global flag keep_orig_idx, no matter it's true for false. - If keep_orig_idx_per_feature is None, fallback to keep_orig_idx control. Note: Adding additional parameter keep_orig_idx_per_feature, instead of change keep_orig_idx directly, is to avoid backward compatibility issue. Differential Revision: D73606958
emlin
added a commit
to emlin/FBGEMM
that referenced
this pull request
May 7, 2025
…tures kernel (pytorch#4027) Summary: X-link: facebookresearch/FBGEMM#1112 **Context** Enhance block_bucketize_sparse_features and block_bucketize_sparse_features_inference kernels to support mixed-format embedding tables. Previously, the keep_orig_idx parameter was a boolean flag applied uniformly across all features, determining whether to retain the original index. With the introduction of [the Flexible Collision-Free Embedding Table](https://github.com/pytorch/torchrec/blob/main/rfc/RFC-0002-Flexible-Collision-Free-Embedding-Table.md), one embedding collection may include both collision-free and collision tables. This update allows the kernel to handle mixed formats by supporting feature-wise control over index retention. For collision-free tables, a large table size of 2^50 is set, maintaining parameters as id-value pairs and preserving the original global id. This change facilitates the use of mixed-style embedding tables effectively. Spec: - keep_orig_idx_per_feature is an optional parameter with per feature settings. - If the keep_orig_idx_per_feature is not None, the value will override global flag keep_orig_idx, no matter it's true for false. - If keep_orig_idx_per_feature is None, fallback to keep_orig_idx control. Note: Adding additional parameter keep_orig_idx_per_feature, instead of change keep_orig_idx directly, is to avoid backward compatibility issue. Differential Revision: D73606958
|
This pull request was exported from Phabricator. Differential Revision: D73606958 |
…tures kernel (pytorch#4027) Summary: X-link: facebookresearch/FBGEMM#1112 **Context** Enhance block_bucketize_sparse_features and block_bucketize_sparse_features_inference kernels to support mixed-format embedding tables. Previously, the keep_orig_idx parameter was a boolean flag applied uniformly across all features, determining whether to retain the original index. With the introduction of [the Flexible Collision-Free Embedding Table](https://github.com/pytorch/torchrec/blob/main/rfc/RFC-0002-Flexible-Collision-Free-Embedding-Table.md), one embedding collection may include both collision-free and collision tables. This update allows the kernel to handle mixed formats by supporting feature-wise control over index retention. For collision-free tables, a large table size of 2^50 is set, maintaining parameters as id-value pairs and preserving the original global id. This change facilitates the use of mixed-style embedding tables effectively. Spec: - keep_orig_idx_per_feature is an optional parameter with per feature settings. - If the keep_orig_idx_per_feature is not None, the value will override global flag keep_orig_idx, no matter it's true for false. - If keep_orig_idx_per_feature is None, fallback to keep_orig_idx control. Note: Adding additional parameter keep_orig_idx_per_feature, instead of change keep_orig_idx directly, is to avoid backward compatibility issue. Differential Revision: D73606958
|
This pull request was exported from Phabricator. Differential Revision: D73606958 |
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:
X-link: https://github.com/facebookresearch/FBGEMM/pull/1112
Context
Enhance block_bucketize_sparse_features and block_bucketize_sparse_features_inference kernels to support mixed-format embedding tables.
Previously, the keep_orig_idx parameter was a boolean flag applied uniformly across all features, determining whether to retain the original index. With the introduction of the Flexible Collision-Free Embedding Table, one embedding collection may include both collision-free and collision tables. This update allows the kernel to handle mixed formats by supporting feature-wise control over index retention.
For collision-free tables, a large table size of 2^50 is set, maintaining parameters as id-value pairs and preserving the original global id. This change facilitates the use of mixed-style embedding tables effectively.
Spec:
Note:
Adding additional parameter keep_orig_idx_per_feature, instead of change keep_orig_idx directly, is to avoid backward compatibility issue.
Differential Revision: D73606958