You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
## Description
GenAI tests are moved to transformers 5.0
Some tests remain on transformers 4.55.4 - `mark.transformers_lower_v5 `
Some tests were divided depending on the cases by models or by
sampling(there are fails with goup beam search) -
`mark.transformers_dependent`
model for tests/python_tests/samples/test_benchmark_vlm.py,
tests/python_tests/samples/test_scheduler_config.py was changed from
tiny-random-minicpmv-2_6 to tiny-random-phi3-vision since support of
minicpmv are deprecating in transformers v.5
(huggingface/optimum-intel#1589)
Test schema example with new marks:
VLM: transformers_dependent(scope for transformers==5.0) + **not**
transformers_lower_v5 + all other tests
VLM with transformers under 5.0: transformers_dependent(scope for
transformers<5.0) + transformers_lower_v5 + **not** all other tests
Problems for transformers_lower_v5 (not all required tasks):
- CVS-185784 whisper tests, error: KeyError: 'num_frames'
- CVS-185787 whisper tests, error: Check '!self_attn_nodes.empty()'
fails
- CVS-185791 accuracy drop llama
- CVS-185790 - group beam search win/lin
- CVS-185788 accuracy drop for tests with Qwen/Qwen2-0.5B-Instruct
- CVS-186071 test_image_generation_cpu_vs_npuw_cpu_with_blob_model fails
- CVS-186059 (CVS-185793 also, need to close) gemma3, llava-next, llava,
error - Add node with name
'__module.model.model.vision_tower.vision_model.embeddings/aten::add/Add'
Check 'input_shape[j] == 1' failed
- llava-next-video hasn't supported by optimum-intel 423b423 with
transformers>=5.0 yet
- version of the speechbrain module compatible with transformers v5.0
causes an import error with k2 module on Windows
Problems for transformers_dependent(the same +):
- CVS-185559 - samples llm tests, conversion of microsoft/phi-1_5 ,
error - AttributeError: 'PhiConfig' object has no attribute
'pad_token_id'. Did you mean: 'bos_token_id'?
- CVS-185605 - "Tokenizers for draft and main models are different" with
optimum-intel 423b423 and transformers>=5.0
- qwen3_next is not supported by optimum-intel 423b423 with
transformers>=5.0
- minicpmv, internvl_chat, minicpmo, llava_next_video is not supported
by transformers>=v5
Task:
[CVS-185024](https://jira.devtools.intel.com/browse/CVS-185024)
## Checklist:
- [ ] This PR follows [GenAI Contributing
guidelines](https://github.com/openvinotoolkit/openvino.genai?tab=contributing-ov-file#contributing).
<!-- Always follow them. If there are deviations, explain what and why.
-->
- [ ] Tests have been updated or added to cover the new code. <!--
Specify exactly which tests were added or updated. If the change isn't
maintenance related, update the tests at
https://github.com/openvinotoolkit/openvino.genai/tree/master/tests or
explain in the description why the tests don't need an update. -->
- [ ] This PR fully addresses the ticket. <!--- If not, explain clearly
what is covered and what is not. If follow-up pull requests are needed,
specify in the description. -->
- [ ] I have made corresponding changes to the documentation. <!-- Run
github.com/\<username>/openvino.genai/actions/workflows/deploy_gh_pages.yml
on your fork with your branch as a parameter to deploy a test version
with the updated content. Replace this comment with the link to the
built docs. If the documentation is updated in a separate PR, clearly
specify it. -->
Copy file name to clipboardExpand all lines: .github/workflows/linux.yml
+68-9Lines changed: 68 additions & 9 deletions
Original file line number
Diff line number
Diff line change
@@ -574,7 +574,13 @@ jobs:
574
574
test:
575
575
- name: 'Whisper'
576
576
# TODO: skip some tests temporary untill https://github.com/huggingface/datasets/issues/7647 dataset is fixed
577
-
cmd: 'python -m pytest -v ./tests/python_tests/test_whisper_pipeline.py ./tests/python_tests/test_whisper_pipeline_static.py -k "not test_smoke[sample_from_dataset0 and not test_whisper_constructors[sample_from_dataset0 and not test_max_new_tokens[sample_from_dataset0 and not test_language_mode[language and not test_task_mode[sample_from_dataset0 and not test_language_autodetect[sample_from_dataset0 and not test_whisper_config_constructor and not test_language_autodetect[sample_from_dataset1 and not test_language_autodetect[sample_from_dataset2 and not test_initial_prompt_hotwords[sample_from_dataset0 and not test_random_sampling[sample_from_dataset0"'
577
+
cmd: 'python -m pytest -v ./tests/python_tests/test_whisper_pipeline.py ./tests/python_tests/test_whisper_pipeline_static.py -m "not transformers_lower_v5" -k "not test_smoke[sample_from_dataset0 and not test_whisper_constructors[sample_from_dataset0 and not test_max_new_tokens[sample_from_dataset0 and not test_language_mode[language and not test_task_mode[sample_from_dataset0 and not test_language_autodetect[sample_from_dataset0 and not test_whisper_config_constructor and not test_language_autodetect[sample_from_dataset1 and not test_language_autodetect[sample_from_dataset2 and not test_initial_prompt_hotwords[sample_from_dataset0 and not test_random_sampling[sample_from_dataset0"'
LD_LIBRARY_PATH: "${{ env.INSTALL_DIR }}/runtime/lib/intel64:${{ env.INSTALL_DIR }}/runtime/3rdparty/tbb/lib:$LD_LIBRARY_PATH"# Required for C++ samples
Copy file name to clipboardExpand all lines: .github/workflows/mac.yml
+23-3Lines changed: 23 additions & 3 deletions
Original file line number
Diff line number
Diff line change
@@ -432,7 +432,14 @@ jobs:
432
432
test:
433
433
- name: 'Whisper'
434
434
# TODO: skip some tests temporary until https://github.com/huggingface/datasets/issues/7647 dataset is fixed
435
-
cmd: 'python -m pytest -v ./tests/python_tests/test_whisper_pipeline.py -k "not test_smoke[sample_from_dataset0 and not test_whisper_constructors[sample_from_dataset0 and not test_max_new_tokens[sample_from_dataset0 and not test_language_mode[language and not test_task_mode[sample_from_dataset0 and not test_language_autodetect[sample_from_dataset0 and not test_whisper_config_constructor and not test_language_autodetect[sample_from_dataset1 and not test_language_autodetect[sample_from_dataset2 and not test_initial_prompt_hotwords[sample_from_dataset0 and not test_random_sampling[sample_from_dataset0"'
435
+
cmd: 'python -m pytest -v ./tests/python_tests/test_whisper_pipeline.py -m "not transformers_lower_v5" -k "not test_smoke[sample_from_dataset0 and not test_whisper_constructors[sample_from_dataset0 and not test_max_new_tokens[sample_from_dataset0 and not test_language_mode[language and not test_task_mode[sample_from_dataset0 and not test_language_autodetect[sample_from_dataset0 and not test_whisper_config_constructor and not test_language_autodetect[sample_from_dataset1 and not test_language_autodetect[sample_from_dataset2 and not test_initial_prompt_hotwords[sample_from_dataset0 and not test_random_sampling[sample_from_dataset0"'
python -m pytest -v ./tests/python_tests/test_whisper_pipeline.py -m "transformers_lower_v5" -k "not test_smoke[sample_from_dataset0 and not test_whisper_constructors[sample_from_dataset0 and not test_max_new_tokens[sample_from_dataset0 and not test_language_mode[language and not test_task_mode[sample_from_dataset0 and not test_language_autodetect[sample_from_dataset0 and not test_whisper_config_constructor and not test_language_autodetect[sample_from_dataset1 and not test_language_autodetect[sample_from_dataset2 and not test_initial_prompt_hotwords[sample_from_dataset0 and not test_random_sampling[sample_from_dataset0"
0 commit comments