[ExecuTorch][WebGPU] Add HuggingFace rotate-half RoPE by JCNTH · Pull Request #21135 · pytorch/executorch

JCNTH · 2026-07-22T04:11:32Z

Stack from ghstack (oldest at bottom):

Add HuggingFace rotate-half RoPE and migrate RoPE to shared dispatch construction

Qwen models pair the first and second halves of each head vector rather than adjacent elements. This adds the rotate-half operator with dynamic start-position updates, full-dimension Q/K handling, generated WGSL, and strict frequency-table bounds.

The shared RoPE handler also closes the dispatch-boilerplate review: it uses graph.device(), typed graph-owned parameter buffers, descriptor-driven bindings and pipelines, named validation and resize callbacks, generated shader-registry lookup, scoped ownership, and graph-owned workgroup recomputation. Interleaved shader math and output are unchanged.

Key changes:

rotary_embedding_hf.wgsl and generated registry entry — one thread per pair for HuggingFace rotate-half.
RotaryEmbedding.cpp — named validation, typed resize contexts, shared dispatch descriptors, and one initial/resize grid picker per route.
Native tests — malformed input rejection, dynamic bounds, shrink/regrow reuse, and HF/interleaved lifecycle coverage.

Co-authored-with: Claude Code.
@exported-using-ghexport

Differential Revision: D113171746

[ghstack-poisoned]

pytorch-bot · 2026-07-22T04:11:35Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/21135

📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 2 New Failures, 29 Pending

As of commit 2372deb with merge base 28a7fac ():

NEW FAILURES - The following jobs have failed:

Cadence Build & Test / cpu-test / test-aot / test-aot (gh)
##[error]API rate limit exceeded for installation. If you reach out to GitHub Support for help, please include the request ID A6A0:E0E81:3793C26:BCF63CC:6A760B4A and timestamp 2026-08-07 16:43:54 UTC. For more on scraping GitHub and how it may affect your rights, please review our Terms of Service (https://docs.github.com/en/site-policy/github-terms/github-terms-of-service) - https://docs.github.com/en/rest/using-the-rest-api/getting-started-with-the-rest-api#rate-limiting
Cadence Build & Test / cpu-test / test-ops / test-ops (gh)
##[error]API rate limit exceeded for installation. If you reach out to GitHub Support for help, please include the request ID ECAC:30BA9A:3991E73:C496DF7:6A760B49 and timestamp 2026-08-07 16:43:53 UTC. For more on scraping GitHub and how it may affect your rights, please review our Terms of Service (https://docs.github.com/en/site-policy/github-terms/github-terms-of-service) - https://docs.github.com/en/rest/using-the-rest-api/getting-started-with-the-rest-api#rate-limiting

This comment was automatically generated by Dr. CI and updates every 15 minutes.

github-actions · 2026-07-22T04:12:22Z

This PR needs a `release notes:` label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

[ghstack-poisoned]

SS-JIA

Review automatically exported from Phabricator review in Meta.

[ghstack-poisoned]

fbsource master [ghstack-poisoned]

Pull Request resolved: #21135 **Add HuggingFace rotate-half RoPE and migrate RoPE to shared dispatch construction** Qwen models pair the first and second halves of each head vector rather than adjacent elements. This adds the rotate-half operator with dynamic start-position updates, full-dimension Q/K handling, generated WGSL, and strict frequency-table bounds. The shared RoPE handler also closes the dispatch-boilerplate review: it uses `graph.device()`, typed graph-owned parameter buffers, descriptor-driven bindings and pipelines, named validation and resize callbacks, generated shader-registry lookup, scoped ownership, and graph-owned workgroup recomputation. Interleaved shader math and output are unchanged. Key changes: - `rotary_embedding_hf.wgsl` and generated registry entry — one thread per pair for HuggingFace rotate-half. - `RotaryEmbedding.cpp` — named validation, typed resize contexts, shared dispatch descriptors, and one initial/resize grid picker per route. - Native tests — malformed input rejection, dynamic bounds, shrink/regrow reuse, and HF/interleaved lifecycle coverage. Co-authored-with: Claude Code. ghstack-source-id: 411961455 @exported-using-ghexport Differential Revision: [D113171746](https://our.internmc.facebook.com/intern/diff/D113171746/)

JCNTH · 2026-07-22T04:11:32Z

Stack from ghstack (oldest at bottom):

Add HuggingFace rotate-half RoPE and migrate RoPE to shared dispatch construction

Qwen models pair the first and second halves of each head vector rather than adjacent elements. This adds the rotate-half operator with dynamic start-position updates, full-dimension Q/K handling, generated WGSL, and strict frequency-table bounds.

The shared RoPE handler also closes the dispatch-boilerplate review: it uses graph.device(), typed graph-owned parameter buffers, descriptor-driven bindings and pipelines, named validation and resize callbacks, generated shader-registry lookup, scoped ownership, and graph-owned workgroup recomputation. Interleaved shader math and output are unchanged.

Key changes:

rotary_embedding_hf.wgsl and generated registry entry — one thread per pair for HuggingFace rotate-half.
RotaryEmbedding.cpp — named validation, typed resize contexts, shared dispatch descriptors, and one initial/resize grid picker per route.
Native tests — malformed input rejection, dynamic bounds, shrink/regrow reuse, and HF/interleaved lifecycle coverage.

Co-authored-with: Claude Code.
@exported-using-ghexport

Differential Revision: D113171746

[ghstack-poisoned]

pytorch-bot · 2026-07-22T04:11:35Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/21135

📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 2 New Failures, 29 Pending

As of commit 2372deb with merge base 28a7fac ():

NEW FAILURES - The following jobs have failed:

Cadence Build & Test / cpu-test / test-aot / test-aot (gh)
##[error]API rate limit exceeded for installation. If you reach out to GitHub Support for help, please include the request ID A6A0:E0E81:3793C26:BCF63CC:6A760B4A and timestamp 2026-08-07 16:43:54 UTC. For more on scraping GitHub and how it may affect your rights, please review our Terms of Service (https://docs.github.com/en/site-policy/github-terms/github-terms-of-service) - https://docs.github.com/en/rest/using-the-rest-api/getting-started-with-the-rest-api#rate-limiting
Cadence Build & Test / cpu-test / test-ops / test-ops (gh)
##[error]API rate limit exceeded for installation. If you reach out to GitHub Support for help, please include the request ID ECAC:30BA9A:3991E73:C496DF7:6A760B49 and timestamp 2026-08-07 16:43:53 UTC. For more on scraping GitHub and how it may affect your rights, please review our Terms of Service (https://docs.github.com/en/site-policy/github-terms/github-terms-of-service) - https://docs.github.com/en/rest/using-the-rest-api/getting-started-with-the-rest-api#rate-limiting

This comment was automatically generated by Dr. CI and updates every 15 minutes.

github-actions · 2026-07-22T04:12:22Z

This PR needs a `release notes:` label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

[ghstack-poisoned]

SS-JIA

Review automatically exported from Phabricator review in Meta.

[ghstack-poisoned]

fbsource master [ghstack-poisoned]

Pull Request resolved: #21135 **Add HuggingFace rotate-half RoPE and migrate RoPE to shared dispatch construction** Qwen models pair the first and second halves of each head vector rather than adjacent elements. This adds the rotate-half operator with dynamic start-position updates, full-dimension Q/K handling, generated WGSL, and strict frequency-table bounds. The shared RoPE handler also closes the dispatch-boilerplate review: it uses `graph.device()`, typed graph-owned parameter buffers, descriptor-driven bindings and pipelines, named validation and resize callbacks, generated shader-registry lookup, scoped ownership, and graph-owned workgroup recomputation. Interleaved shader math and output are unchanged. Key changes: - `rotary_embedding_hf.wgsl` and generated registry entry — one thread per pair for HuggingFace rotate-half. - `RotaryEmbedding.cpp` — named validation, typed resize contexts, shared dispatch descriptors, and one initial/resize grid picker per route. - Native tests — malformed input rejection, dynamic bounds, shrink/regrow reuse, and HF/interleaved lifecycle coverage. Co-authored-with: Claude Code. ghstack-source-id: 411961455 @exported-using-ghexport Differential Revision: [D113171746](https://our.internmc.facebook.com/intern/diff/D113171746/)

JCNTH · 2026-07-22T04:11:32Z

Stack from ghstack (oldest at bottom):

Add HuggingFace rotate-half RoPE and migrate RoPE to shared dispatch construction

Qwen models pair the first and second halves of each head vector rather than adjacent elements. This adds the rotate-half operator with dynamic start-position updates, full-dimension Q/K handling, generated WGSL, and strict frequency-table bounds.

The shared RoPE handler also closes the dispatch-boilerplate review: it uses graph.device(), typed graph-owned parameter buffers, descriptor-driven bindings and pipelines, named validation and resize callbacks, generated shader-registry lookup, scoped ownership, and graph-owned workgroup recomputation. Interleaved shader math and output are unchanged.

Key changes:

rotary_embedding_hf.wgsl and generated registry entry — one thread per pair for HuggingFace rotate-half.
RotaryEmbedding.cpp — named validation, typed resize contexts, shared dispatch descriptors, and one initial/resize grid picker per route.
Native tests — malformed input rejection, dynamic bounds, shrink/regrow reuse, and HF/interleaved lifecycle coverage.

Co-authored-with: Claude Code.
@exported-using-ghexport

Differential Revision: D113171746

[ghstack-poisoned]

pytorch-bot · 2026-07-22T04:11:35Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/21135

📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 2 New Failures, 29 Pending

As of commit 2372deb with merge base 28a7fac ():

NEW FAILURES - The following jobs have failed:

Cadence Build & Test / cpu-test / test-aot / test-aot (gh)
##[error]API rate limit exceeded for installation. If you reach out to GitHub Support for help, please include the request ID A6A0:E0E81:3793C26:BCF63CC:6A760B4A and timestamp 2026-08-07 16:43:54 UTC. For more on scraping GitHub and how it may affect your rights, please review our Terms of Service (https://docs.github.com/en/site-policy/github-terms/github-terms-of-service) - https://docs.github.com/en/rest/using-the-rest-api/getting-started-with-the-rest-api#rate-limiting
Cadence Build & Test / cpu-test / test-ops / test-ops (gh)
##[error]API rate limit exceeded for installation. If you reach out to GitHub Support for help, please include the request ID ECAC:30BA9A:3991E73:C496DF7:6A760B49 and timestamp 2026-08-07 16:43:53 UTC. For more on scraping GitHub and how it may affect your rights, please review our Terms of Service (https://docs.github.com/en/site-policy/github-terms/github-terms-of-service) - https://docs.github.com/en/rest/using-the-rest-api/getting-started-with-the-rest-api#rate-limiting

This comment was automatically generated by Dr. CI and updates every 15 minutes.

github-actions · 2026-07-22T04:12:22Z

This PR needs a `release notes:` label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

[ghstack-poisoned]

SS-JIA

Review automatically exported from Phabricator review in Meta.

[ghstack-poisoned]

fbsource master [ghstack-poisoned]

Pull Request resolved: #21135 **Add HuggingFace rotate-half RoPE and migrate RoPE to shared dispatch construction** Qwen models pair the first and second halves of each head vector rather than adjacent elements. This adds the rotate-half operator with dynamic start-position updates, full-dimension Q/K handling, generated WGSL, and strict frequency-table bounds. The shared RoPE handler also closes the dispatch-boilerplate review: it uses `graph.device()`, typed graph-owned parameter buffers, descriptor-driven bindings and pipelines, named validation and resize callbacks, generated shader-registry lookup, scoped ownership, and graph-owned workgroup recomputation. Interleaved shader math and output are unchanged. Key changes: - `rotary_embedding_hf.wgsl` and generated registry entry — one thread per pair for HuggingFace rotate-half. - `RotaryEmbedding.cpp` — named validation, typed resize contexts, shared dispatch descriptors, and one initial/resize grid picker per route. - Native tests — malformed input rejection, dynamic bounds, shrink/regrow reuse, and HF/interleaved lifecycle coverage. Co-authored-with: Claude Code. ghstack-source-id: 411961455 @exported-using-ghexport Differential Revision: [D113171746](https://our.internmc.facebook.com/intern/diff/D113171746/)

JCNTH · 2026-07-22T04:11:32Z

Stack from ghstack (oldest at bottom):

Add HuggingFace rotate-half RoPE and migrate RoPE to shared dispatch construction

Qwen models pair the first and second halves of each head vector rather than adjacent elements. This adds the rotate-half operator with dynamic start-position updates, full-dimension Q/K handling, generated WGSL, and strict frequency-table bounds.

The shared RoPE handler also closes the dispatch-boilerplate review: it uses graph.device(), typed graph-owned parameter buffers, descriptor-driven bindings and pipelines, named validation and resize callbacks, generated shader-registry lookup, scoped ownership, and graph-owned workgroup recomputation. Interleaved shader math and output are unchanged.

Key changes:

rotary_embedding_hf.wgsl and generated registry entry — one thread per pair for HuggingFace rotate-half.
RotaryEmbedding.cpp — named validation, typed resize contexts, shared dispatch descriptors, and one initial/resize grid picker per route.
Native tests — malformed input rejection, dynamic bounds, shrink/regrow reuse, and HF/interleaved lifecycle coverage.

Co-authored-with: Claude Code.
@exported-using-ghexport

Differential Revision: D113171746

[ghstack-poisoned]

pytorch-bot · 2026-07-22T04:11:35Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/21135

📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 2 New Failures, 29 Pending

As of commit 2372deb with merge base 28a7fac ():

NEW FAILURES - The following jobs have failed:

Cadence Build & Test / cpu-test / test-aot / test-aot (gh)
##[error]API rate limit exceeded for installation. If you reach out to GitHub Support for help, please include the request ID A6A0:E0E81:3793C26:BCF63CC:6A760B4A and timestamp 2026-08-07 16:43:54 UTC. For more on scraping GitHub and how it may affect your rights, please review our Terms of Service (https://docs.github.com/en/site-policy/github-terms/github-terms-of-service) - https://docs.github.com/en/rest/using-the-rest-api/getting-started-with-the-rest-api#rate-limiting
Cadence Build & Test / cpu-test / test-ops / test-ops (gh)
##[error]API rate limit exceeded for installation. If you reach out to GitHub Support for help, please include the request ID ECAC:30BA9A:3991E73:C496DF7:6A760B49 and timestamp 2026-08-07 16:43:53 UTC. For more on scraping GitHub and how it may affect your rights, please review our Terms of Service (https://docs.github.com/en/site-policy/github-terms/github-terms-of-service) - https://docs.github.com/en/rest/using-the-rest-api/getting-started-with-the-rest-api#rate-limiting

This comment was automatically generated by Dr. CI and updates every 15 minutes.

github-actions · 2026-07-22T04:12:22Z

This PR needs a `release notes:` label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

[ghstack-poisoned]

SS-JIA

Review automatically exported from Phabricator review in Meta.

[ghstack-poisoned]

fbsource master [ghstack-poisoned]

Pull Request resolved: #21135 **Add HuggingFace rotate-half RoPE and migrate RoPE to shared dispatch construction** Qwen models pair the first and second halves of each head vector rather than adjacent elements. This adds the rotate-half operator with dynamic start-position updates, full-dimension Q/K handling, generated WGSL, and strict frequency-table bounds. The shared RoPE handler also closes the dispatch-boilerplate review: it uses `graph.device()`, typed graph-owned parameter buffers, descriptor-driven bindings and pipelines, named validation and resize callbacks, generated shader-registry lookup, scoped ownership, and graph-owned workgroup recomputation. Interleaved shader math and output are unchanged. Key changes: - `rotary_embedding_hf.wgsl` and generated registry entry — one thread per pair for HuggingFace rotate-half. - `RotaryEmbedding.cpp` — named validation, typed resize contexts, shared dispatch descriptors, and one initial/resize grid picker per route. - Native tests — malformed input rejection, dynamic bounds, shrink/regrow reuse, and HF/interleaved lifecycle coverage. Co-authored-with: Claude Code. ghstack-source-id: 411961455 @exported-using-ghexport Differential Revision: [D113171746](https://our.internmc.facebook.com/intern/diff/D113171746/)

JCNTH · 2026-07-22T04:11:32Z

Stack from ghstack (oldest at bottom):

Add HuggingFace rotate-half RoPE and migrate RoPE to shared dispatch construction

Qwen models pair the first and second halves of each head vector rather than adjacent elements. This adds the rotate-half operator with dynamic start-position updates, full-dimension Q/K handling, generated WGSL, and strict frequency-table bounds.

The shared RoPE handler also closes the dispatch-boilerplate review: it uses graph.device(), typed graph-owned parameter buffers, descriptor-driven bindings and pipelines, named validation and resize callbacks, generated shader-registry lookup, scoped ownership, and graph-owned workgroup recomputation. Interleaved shader math and output are unchanged.

Key changes:

rotary_embedding_hf.wgsl and generated registry entry — one thread per pair for HuggingFace rotate-half.
RotaryEmbedding.cpp — named validation, typed resize contexts, shared dispatch descriptors, and one initial/resize grid picker per route.
Native tests — malformed input rejection, dynamic bounds, shrink/regrow reuse, and HF/interleaved lifecycle coverage.

Co-authored-with: Claude Code.
@exported-using-ghexport

Differential Revision: D113171746

[ghstack-poisoned]

pytorch-bot · 2026-07-22T04:11:35Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/21135

📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 2 New Failures, 29 Pending

As of commit 2372deb with merge base 28a7fac ():

NEW FAILURES - The following jobs have failed:

Cadence Build & Test / cpu-test / test-aot / test-aot (gh)
##[error]API rate limit exceeded for installation. If you reach out to GitHub Support for help, please include the request ID A6A0:E0E81:3793C26:BCF63CC:6A760B4A and timestamp 2026-08-07 16:43:54 UTC. For more on scraping GitHub and how it may affect your rights, please review our Terms of Service (https://docs.github.com/en/site-policy/github-terms/github-terms-of-service) - https://docs.github.com/en/rest/using-the-rest-api/getting-started-with-the-rest-api#rate-limiting
Cadence Build & Test / cpu-test / test-ops / test-ops (gh)
##[error]API rate limit exceeded for installation. If you reach out to GitHub Support for help, please include the request ID ECAC:30BA9A:3991E73:C496DF7:6A760B49 and timestamp 2026-08-07 16:43:53 UTC. For more on scraping GitHub and how it may affect your rights, please review our Terms of Service (https://docs.github.com/en/site-policy/github-terms/github-terms-of-service) - https://docs.github.com/en/rest/using-the-rest-api/getting-started-with-the-rest-api#rate-limiting

This comment was automatically generated by Dr. CI and updates every 15 minutes.

github-actions · 2026-07-22T04:12:22Z

This PR needs a `release notes:` label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

[ghstack-poisoned]

SS-JIA

Review automatically exported from Phabricator review in Meta.

[ghstack-poisoned]

fbsource master [ghstack-poisoned]

Pull Request resolved: #21135 **Add HuggingFace rotate-half RoPE and migrate RoPE to shared dispatch construction** Qwen models pair the first and second halves of each head vector rather than adjacent elements. This adds the rotate-half operator with dynamic start-position updates, full-dimension Q/K handling, generated WGSL, and strict frequency-table bounds. The shared RoPE handler also closes the dispatch-boilerplate review: it uses `graph.device()`, typed graph-owned parameter buffers, descriptor-driven bindings and pipelines, named validation and resize callbacks, generated shader-registry lookup, scoped ownership, and graph-owned workgroup recomputation. Interleaved shader math and output are unchanged. Key changes: - `rotary_embedding_hf.wgsl` and generated registry entry — one thread per pair for HuggingFace rotate-half. - `RotaryEmbedding.cpp` — named validation, typed resize contexts, shared dispatch descriptors, and one initial/resize grid picker per route. - Native tests — malformed input rejection, dynamic bounds, shrink/regrow reuse, and HF/interleaved lifecycle coverage. Co-authored-with: Claude Code. ghstack-source-id: 411961455 @exported-using-ghexport Differential Revision: [D113171746](https://our.internmc.facebook.com/intern/diff/D113171746/)

JCNTH · 2026-07-22T04:11:32Z

Stack from ghstack (oldest at bottom):

Add HuggingFace rotate-half RoPE and migrate RoPE to shared dispatch construction

Qwen models pair the first and second halves of each head vector rather than adjacent elements. This adds the rotate-half operator with dynamic start-position updates, full-dimension Q/K handling, generated WGSL, and strict frequency-table bounds.

The shared RoPE handler also closes the dispatch-boilerplate review: it uses graph.device(), typed graph-owned parameter buffers, descriptor-driven bindings and pipelines, named validation and resize callbacks, generated shader-registry lookup, scoped ownership, and graph-owned workgroup recomputation. Interleaved shader math and output are unchanged.

Key changes:

rotary_embedding_hf.wgsl and generated registry entry — one thread per pair for HuggingFace rotate-half.
RotaryEmbedding.cpp — named validation, typed resize contexts, shared dispatch descriptors, and one initial/resize grid picker per route.
Native tests — malformed input rejection, dynamic bounds, shrink/regrow reuse, and HF/interleaved lifecycle coverage.

Co-authored-with: Claude Code.
@exported-using-ghexport

Differential Revision: D113171746

[ghstack-poisoned]

pytorch-bot · 2026-07-22T04:11:35Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/21135

📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 2 New Failures, 29 Pending

As of commit 2372deb with merge base 28a7fac ():

NEW FAILURES - The following jobs have failed:

Cadence Build & Test / cpu-test / test-aot / test-aot (gh)
##[error]API rate limit exceeded for installation. If you reach out to GitHub Support for help, please include the request ID A6A0:E0E81:3793C26:BCF63CC:6A760B4A and timestamp 2026-08-07 16:43:54 UTC. For more on scraping GitHub and how it may affect your rights, please review our Terms of Service (https://docs.github.com/en/site-policy/github-terms/github-terms-of-service) - https://docs.github.com/en/rest/using-the-rest-api/getting-started-with-the-rest-api#rate-limiting
Cadence Build & Test / cpu-test / test-ops / test-ops (gh)
##[error]API rate limit exceeded for installation. If you reach out to GitHub Support for help, please include the request ID ECAC:30BA9A:3991E73:C496DF7:6A760B49 and timestamp 2026-08-07 16:43:53 UTC. For more on scraping GitHub and how it may affect your rights, please review our Terms of Service (https://docs.github.com/en/site-policy/github-terms/github-terms-of-service) - https://docs.github.com/en/rest/using-the-rest-api/getting-started-with-the-rest-api#rate-limiting

This comment was automatically generated by Dr. CI and updates every 15 minutes.

github-actions · 2026-07-22T04:12:22Z

This PR needs a `release notes:` label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

[ghstack-poisoned]

SS-JIA

Review automatically exported from Phabricator review in Meta.

[ghstack-poisoned]

fbsource master [ghstack-poisoned]

Pull Request resolved: #21135 **Add HuggingFace rotate-half RoPE and migrate RoPE to shared dispatch construction** Qwen models pair the first and second halves of each head vector rather than adjacent elements. This adds the rotate-half operator with dynamic start-position updates, full-dimension Q/K handling, generated WGSL, and strict frequency-table bounds. The shared RoPE handler also closes the dispatch-boilerplate review: it uses `graph.device()`, typed graph-owned parameter buffers, descriptor-driven bindings and pipelines, named validation and resize callbacks, generated shader-registry lookup, scoped ownership, and graph-owned workgroup recomputation. Interleaved shader math and output are unchanged. Key changes: - `rotary_embedding_hf.wgsl` and generated registry entry — one thread per pair for HuggingFace rotate-half. - `RotaryEmbedding.cpp` — named validation, typed resize contexts, shared dispatch descriptors, and one initial/resize grid picker per route. - Native tests — malformed input rejection, dynamic bounds, shrink/regrow reuse, and HF/interleaved lifecycle coverage. Co-authored-with: Claude Code. ghstack-source-id: 411961455 @exported-using-ghexport Differential Revision: [D113171746](https://our.internmc.facebook.com/intern/diff/D113171746/)

JCNTH · 2026-07-22T04:11:32Z

Stack from ghstack (oldest at bottom):

Add HuggingFace rotate-half RoPE and migrate RoPE to shared dispatch construction

Qwen models pair the first and second halves of each head vector rather than adjacent elements. This adds the rotate-half operator with dynamic start-position updates, full-dimension Q/K handling, generated WGSL, and strict frequency-table bounds.

The shared RoPE handler also closes the dispatch-boilerplate review: it uses graph.device(), typed graph-owned parameter buffers, descriptor-driven bindings and pipelines, named validation and resize callbacks, generated shader-registry lookup, scoped ownership, and graph-owned workgroup recomputation. Interleaved shader math and output are unchanged.

Key changes:

rotary_embedding_hf.wgsl and generated registry entry — one thread per pair for HuggingFace rotate-half.
RotaryEmbedding.cpp — named validation, typed resize contexts, shared dispatch descriptors, and one initial/resize grid picker per route.
Native tests — malformed input rejection, dynamic bounds, shrink/regrow reuse, and HF/interleaved lifecycle coverage.

Co-authored-with: Claude Code.
@exported-using-ghexport

Differential Revision: D113171746

[ghstack-poisoned]

pytorch-bot · 2026-07-22T04:11:35Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/21135

📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 2 New Failures, 29 Pending

As of commit 2372deb with merge base 28a7fac ():

NEW FAILURES - The following jobs have failed:

Cadence Build & Test / cpu-test / test-aot / test-aot (gh)
##[error]API rate limit exceeded for installation. If you reach out to GitHub Support for help, please include the request ID A6A0:E0E81:3793C26:BCF63CC:6A760B4A and timestamp 2026-08-07 16:43:54 UTC. For more on scraping GitHub and how it may affect your rights, please review our Terms of Service (https://docs.github.com/en/site-policy/github-terms/github-terms-of-service) - https://docs.github.com/en/rest/using-the-rest-api/getting-started-with-the-rest-api#rate-limiting
Cadence Build & Test / cpu-test / test-ops / test-ops (gh)
##[error]API rate limit exceeded for installation. If you reach out to GitHub Support for help, please include the request ID ECAC:30BA9A:3991E73:C496DF7:6A760B49 and timestamp 2026-08-07 16:43:53 UTC. For more on scraping GitHub and how it may affect your rights, please review our Terms of Service (https://docs.github.com/en/site-policy/github-terms/github-terms-of-service) - https://docs.github.com/en/rest/using-the-rest-api/getting-started-with-the-rest-api#rate-limiting

This comment was automatically generated by Dr. CI and updates every 15 minutes.

github-actions · 2026-07-22T04:12:22Z

This PR needs a `release notes:` label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

[ghstack-poisoned]

SS-JIA

Review automatically exported from Phabricator review in Meta.

[ghstack-poisoned]

fbsource master [ghstack-poisoned]

Pull Request resolved: #21135 **Add HuggingFace rotate-half RoPE and migrate RoPE to shared dispatch construction** Qwen models pair the first and second halves of each head vector rather than adjacent elements. This adds the rotate-half operator with dynamic start-position updates, full-dimension Q/K handling, generated WGSL, and strict frequency-table bounds. The shared RoPE handler also closes the dispatch-boilerplate review: it uses `graph.device()`, typed graph-owned parameter buffers, descriptor-driven bindings and pipelines, named validation and resize callbacks, generated shader-registry lookup, scoped ownership, and graph-owned workgroup recomputation. Interleaved shader math and output are unchanged. Key changes: - `rotary_embedding_hf.wgsl` and generated registry entry — one thread per pair for HuggingFace rotate-half. - `RotaryEmbedding.cpp` — named validation, typed resize contexts, shared dispatch descriptors, and one initial/resize grid picker per route. - Native tests — malformed input rejection, dynamic bounds, shrink/regrow reuse, and HF/interleaved lifecycle coverage. Co-authored-with: Claude Code. ghstack-source-id: 411961455 @exported-using-ghexport Differential Revision: [D113171746](https://our.internmc.facebook.com/intern/diff/D113171746/)

JCNTH · 2026-07-22T04:11:32Z

Stack from ghstack (oldest at bottom):

Add HuggingFace rotate-half RoPE and migrate RoPE to shared dispatch construction

Qwen models pair the first and second halves of each head vector rather than adjacent elements. This adds the rotate-half operator with dynamic start-position updates, full-dimension Q/K handling, generated WGSL, and strict frequency-table bounds.

The shared RoPE handler also closes the dispatch-boilerplate review: it uses graph.device(), typed graph-owned parameter buffers, descriptor-driven bindings and pipelines, named validation and resize callbacks, generated shader-registry lookup, scoped ownership, and graph-owned workgroup recomputation. Interleaved shader math and output are unchanged.

Key changes:

rotary_embedding_hf.wgsl and generated registry entry — one thread per pair for HuggingFace rotate-half.
RotaryEmbedding.cpp — named validation, typed resize contexts, shared dispatch descriptors, and one initial/resize grid picker per route.
Native tests — malformed input rejection, dynamic bounds, shrink/regrow reuse, and HF/interleaved lifecycle coverage.

Co-authored-with: Claude Code.
@exported-using-ghexport

Differential Revision: D113171746

[ghstack-poisoned]

pytorch-bot · 2026-07-22T04:11:35Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/21135

📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 2 New Failures, 29 Pending

As of commit 2372deb with merge base 28a7fac ():

NEW FAILURES - The following jobs have failed:

Cadence Build & Test / cpu-test / test-aot / test-aot (gh)
##[error]API rate limit exceeded for installation. If you reach out to GitHub Support for help, please include the request ID A6A0:E0E81:3793C26:BCF63CC:6A760B4A and timestamp 2026-08-07 16:43:54 UTC. For more on scraping GitHub and how it may affect your rights, please review our Terms of Service (https://docs.github.com/en/site-policy/github-terms/github-terms-of-service) - https://docs.github.com/en/rest/using-the-rest-api/getting-started-with-the-rest-api#rate-limiting
Cadence Build & Test / cpu-test / test-ops / test-ops (gh)
##[error]API rate limit exceeded for installation. If you reach out to GitHub Support for help, please include the request ID ECAC:30BA9A:3991E73:C496DF7:6A760B49 and timestamp 2026-08-07 16:43:53 UTC. For more on scraping GitHub and how it may affect your rights, please review our Terms of Service (https://docs.github.com/en/site-policy/github-terms/github-terms-of-service) - https://docs.github.com/en/rest/using-the-rest-api/getting-started-with-the-rest-api#rate-limiting

This comment was automatically generated by Dr. CI and updates every 15 minutes.

github-actions · 2026-07-22T04:12:22Z

This PR needs a `release notes:` label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

[ghstack-poisoned]

SS-JIA

Review automatically exported from Phabricator review in Meta.

[ghstack-poisoned]

fbsource master [ghstack-poisoned]

Pull Request resolved: #21135 **Add HuggingFace rotate-half RoPE and migrate RoPE to shared dispatch construction** Qwen models pair the first and second halves of each head vector rather than adjacent elements. This adds the rotate-half operator with dynamic start-position updates, full-dimension Q/K handling, generated WGSL, and strict frequency-table bounds. The shared RoPE handler also closes the dispatch-boilerplate review: it uses `graph.device()`, typed graph-owned parameter buffers, descriptor-driven bindings and pipelines, named validation and resize callbacks, generated shader-registry lookup, scoped ownership, and graph-owned workgroup recomputation. Interleaved shader math and output are unchanged. Key changes: - `rotary_embedding_hf.wgsl` and generated registry entry — one thread per pair for HuggingFace rotate-half. - `RotaryEmbedding.cpp` — named validation, typed resize contexts, shared dispatch descriptors, and one initial/resize grid picker per route. - Native tests — malformed input rejection, dynamic bounds, shrink/regrow reuse, and HF/interleaved lifecycle coverage. Co-authored-with: Claude Code. ghstack-source-id: 411961455 @exported-using-ghexport Differential Revision: [D113171746](https://our.internmc.facebook.com/intern/diff/D113171746/)

Update
582b4f9
[ghstack-poisoned]

JCNTH had a problem deploying to cadence July 22, 2026 04:11 — with GitHub Actions Error

meta-claBot added the CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. label Jul 22, 2026

Update
e04a54a
[ghstack-poisoned]

meta-codesyncBot added the meta-exported label Jul 22, 2026

JCNTH temporarily deployed to cadence July 22, 2026 04:24 — with GitHub Actions Inactive

JCNTH temporarily deployed to cadence July 22, 2026 04:49 — with GitHub Actions Inactive

Update
5ef7a22
[ghstack-poisoned]

JCNTH temporarily deployed to cadence July 22, 2026 17:33 — with GitHub Actions Inactive

JCNTH had a problem deploying to cadence July 22, 2026 18:08 — with GitHub Actions Error

Update
0c45a43
[ghstack-poisoned]

JCNTH temporarily deployed to cadence July 22, 2026 18:27 — with GitHub Actions Inactive

JCNTH temporarily deployed to cadence July 22, 2026 18:57 — with GitHub Actions Inactive

SS-JIA approved these changes Jul 22, 2026
View reviewed changes

Update
12c00a5
[ghstack-poisoned]

Update
53506e9
[ghstack-poisoned]

JCNTH temporarily deployed to cadence July 31, 2026 21:21 — with GitHub Actions Inactive

JCNTH temporarily deployed to cadence July 31, 2026 21:52 — with GitHub Actions Inactive

Update
fc0fa3c
[ghstack-poisoned]

JCNTH temporarily deployed to cadence August 5, 2026 19:55 — with GitHub Actions Inactive

JCNTH temporarily deployed to cadence August 5, 2026 21:04 — with GitHub Actions Inactive

Rebase onto latest
2372deb
fbsource master [ghstack-poisoned]

JCNTH temporarily deployed to cadence August 7, 2026 16:36 — with GitHub Actions Inactive

JCNTH temporarily deployed to cadence August 7, 2026 17:00 — with GitHub Actions Inactive

meta-codesyncBot merged commit 956057c into gh/JCNTH/111/baseAug 7, 2026
180 of 183 checks passed

meta-codesyncBot deleted the gh/JCNTH/111/head branch August 7, 2026 17:21

meta-codesyncBottemporarily deployed to cherry-pick-bot August 7, 2026 17:21 Inactive

pytorchbot mentioned this pull request Aug 7, 2026
[ExecuTorch][WebGPU] Add HuggingFace rotate-half RoPE #21654
Merged