[Qwen2.5-7B] Error in PTQ calibration, occured "ImportError: cannot import name VideoInput" when running "bash 1_make_ptq_calibration_dataset.sh"

NeuroPilot NeuroPilot - GenAI
,
1

Referring to “Tutorial for Large Language Models”, I used the Qwen2.5-7B model for the experiment.
When I reached step “5.1.3.4.1. Preparing the Model for PTQ” and ran the command “bash 1_make_ptq_calibration_dataset.sh”, I occured the following error.
Does anyone know what might be causing this?

Traceback (most recent call last):
File “/home/mediatek/miniconda3/envs/qwen/bin/mtk_make_llm_ptq_calib_dataset”, line 7, in
sys.exit(main())
^^^^^^
File “mtk_llm_sdk/utils/memory_profiler.py”, line 281, in mtk_llm_sdk.utils.memory_profiler.memory_peak_profile.wrapper
File “mtk_llm_sdk/utils/memory_profiler.py”, line 282, in mtk_llm_sdk.utils.memory_profiler.memory_peak_profile.wrapper
File “mtk_llm_sdk/make_ptq_calibration_dataset.py”, line 547, in mtk_llm_sdk.make_ptq_calibration_dataset.main
File “mtk_llm_sdk/utils/utils.py”, line 544, in mtk_llm_sdk.utils.utils.instantiate_models
File “mtk_llm_sdk/models/mllm/modeling_qwen2vl_vision.py”, line 18, in init mtk_llm_sdk.models.mllm.modeling_qwen2vl_vision
ImportError: cannot import name VideoInput
2

To resolve this error, upgrade your transformers library to version 4.45 or above:

pip install --upgrade "transformers==4.45.2"

After upgrading, re-run the PTQ calibration dataset script:

bash 1_make_ptq_calibration_dataset.sh

This should ensure the required VideoInput class is present and properly imported.