[EXPERIMENT][WIP] Add Qwen3.5 support to OpenVINO export#1746
Draft
mlukasze wants to merge 1 commit into
Draft
Conversation
Adds export support fixes for Qwen/Qwen3.5-27B and related architectures:\n- fix qwen3_5 dynamic cache mask sizing during traced export\n- extend supported transformers window for qwen3_5/qwen3_5_moe\n- update related docs and OpenVINO test gates\n\nRequires transformers >= 5.2.0\n\nCo-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What does this PR do?
Adds and validates Qwen3.5 / Qwen3.5-MoE OpenVINO export support against modern Transformers snapshots. The patch fixes Qwen3.5 dynamic cache mask sizing during traced export, relaxes the supported Transformers upper bound to cover the current upstream Qwen3.5 implementation, and updates the related OpenVINO tests/docs gates. Validation performed on the tiny internal HF checkpoints:
optimum-cli export openvino --model optimum-intel-internal-testing/tiny-random-qwen3.5 --task image-text-to-text ...now succeeds, OVModelForVisualCausalLM loads the export on CPU, andpytest tests/openvino/test_export.py -k qwen3_5 -x -v --no-headerpasses.Installation instructions
Exporting cmd-line
optimum-cli export openvino --model optimum-intel-internal-testing/tiny-random-qwen3.5 --task image-text-to-text /opt/home/mlukasze/dev/Qwen-Qwen3.5-27B/tiny-qwen35-ov --trust-remote-codeInference script
Before submitting