Skip to content

Update transformers version from 5.0.0rc1 to 5.11.0 - #24422

Merged
Jiseong-oh merged 6 commits into
mainfrom
update-transformers-5.11.0
Nov 9, 2026
Merged

Update transformers version from 5.0.0rc1 to 5.11.0#24422
Jiseong-oh merged 6 commits into
mainfrom
update-transformers-5.11.0

Conversation

@Jiseong-oh

@Jiseong-oh Jiseong-oh commented Nov 2, 2026 •
edited by pytorch-bot Bot
Loading

Copy link
Copy Markdown
Collaborator

For supporting Gemma4-QAT model, it should be used tranformers 5.11.0 version.

cc @digantdesai @freddan80 @per @zingo @oscarandersson8218 @mansnils @Sebastian-Larsson @robell @rascani @SS-JIA @kimishpatel

@Jiseong-oh Jiseong-oh added CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. partner: samsung For backend delegation, kernels, demo, etc. from the 3rd-party partner, Samsung module: samsung labels Nov 2, 2026
@pytorch-bot

pytorch-bot Bot commented Nov 2, 2026 •
edited
Loading

Copy link
Copy Markdown

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/24422

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

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

@github-actions

github-actions Bot commented Nov 2, 2026

Copy link
Copy Markdown

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.

@github-actions github-actions Bot added ciflow/trunk module: arm Issues related to arm backend labels Nov 3, 2026
@Jiseong-oh
Jiseong-oh force-pushed the update-transformers-5.11.0 branch from a91eb79 to 58ce574 Compare November 3, 2026 11:17
@Jiseong-oh
Jiseong-oh requested a review from SS-JIA November 3, 2026 11:20
@Jiseong-oh
Update transformers version from 5.0.0rc1 to 5.11.0
64d7f85 
For supporting Gemma4-QAT model, it should be used
tranformers 5.11.0 version.

Signed-off-by: jiseong.oh <jiseong.oh@samsung.com>
@Jiseong-oh
fix compliation issue
104da5e 
- Update codes according to transformers version

Signed-off-by: jiseong.oh <jiseong.oh@samsung.com>
@Jiseong-oh
Write HF static KV cache rows at the cache_position input
bad9e28 
With transformers 5.11.0, StaticLayer.update ignores cache_position and writes
K/V at arange(n) + cumulative_length. Setting cumulative_length from
cache_position keeps eager results right, but the exported cache writes then
index rows with the output of an add. The ggml engine only accepts index_put_
row indices that come from a graph input or an integer arange, so
updated codes

Signed-off-by: jiseong.oh <jiseong.oh@samsung.com>
@Jiseong-oh
Jiseong-oh force-pushed the update-transformers-5.11.0 branch from 83b4b1f to bad9e28 Compare November 6, 2026 03:15
@shoumikhin

shoumikhin commented Nov 9, 2026

Copy link
Copy Markdown
Contributor

heads up, test-model-metal-e2e has been red on main since this landed, all 6 rows.

whisper: RuntimeError: Constant L__self___cache_self_attention_cache_layers_0_cumulative_length is mutated in the forward method. Pls register it as buffer
voxtral: AttributeError: 'BaseModelOutputWithPooling' object has no attribute 'unsqueeze'

the metal job installs the optimum-executorch pin first (transformers 5.0.0rc1), then install_executorch.sh pulls requirements-examples.txt, which now upgrades transformers to 5.11.0. cuda e2e installs them in the other order, so it still ends on 5.0.0rc1 and stays green. likely fix: in metal.yml install optimum-executorch after install_executorch.sh like cuda.yml does, or bump the optimum-executorch pin to one that works with 5.11

@Jiseong-oh

Jiseong-oh commented Nov 10, 2026

Copy link
Copy Markdown
Collaborator Author

heads up, test-model-metal-e2e has been red on main since this landed, all 6 rows.

whisper: RuntimeError: Constant L__self___cache_self_attention_cache_layers_0_cumulative_length is mutated in the forward method. Pls register it as buffer voxtral: AttributeError: 'BaseModelOutputWithPooling' object has no attribute 'unsqueeze'

the metal job installs the optimum-executorch pin first (transformers 5.0.0rc1), then install_executorch.sh pulls requirements-examples.txt, which now upgrades transformers to 5.11.0. cuda e2e installs them in the other order, so it still ends on 5.0.0rc1 and stays green. likely fix: in metal.yml install optimum-executorch after install_executorch.sh like cuda.yml does, or bump the optimum-executorch pin to one that works with 5.11

Thanks @shoumikhin for the information. I put a pr for fixing the issue

@SS-JIA SS-JIA mentioned this pull request Nov 10, 2026
Merged
@JakeStevens JakeStevens mentioned this pull request Nov 10, 2026
Merged

This branch was previously deployed

1 inactive deployment
cadence — 749347f4 Deployed Nov 9, 2026 by Jiseong-oh via hifi5s-op-test / hifi5s #37331
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@SS-JIA SS-JIA SS-JIA approved these changes
@psiddh psiddh Awaiting requested review from psiddh psiddh is a code owner
@digantdesai digantdesai Awaiting requested review from digantdesai digantdesai is a code owner
@rascani rascani Awaiting requested review from rascani rascani is a code owner

Assignees

No one assigned

Labels

ciflow/trunk CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. module: arm Issues related to arm backend module: samsung partner: samsung For backend delegation, kernels, demo, etc. from the 3rd-party partner, Samsung

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

3 participants

@Jiseong-oh @shoumikhin @SS-JIA