Try importing gptqmodel first when exporting a GPTQ model to openvino

[notsyncing]

What does this PR do?

Hello, I'm trying exporting a Qwen2.5 GPTQ model loaded with gptqmodel to openvino, and it errors:

INFO  ENV: Auto setting PYTORCH_CUDA_ALLOC_CONF='expandable_segments:True' for memory saving.                                                       
INFO  ENV: Auto setting CUDA_DEVICE_ORDER=PCI_BUS_ID for correctness.                                                                               
`low_cpu_mem_usage` was None, now default to True since model is quantized.
Sliding Window Attention is enabled but not implemented for `sdpa`; unexpected results may be encountered.
INFO   Kernel: Auto-selection: adding candidate `IPEXQuantLinear`                                                                                   
`loss_type=None` was set in the config but it is unrecognised.Using the default loss: `ForCausalLMLoss`.
Traceback (most recent call last):
  File "/var/home/sfc/Projects/azarrot-py312/.venv/bin/azarrot", line 10, in <module>
    sys.exit(main())
             ^^^^^^
  File "/var/home/sfc/Projects/azarrot-py312/src/azarrot/server.py", line 347, in main
    server = create_server()
             ^^^^^^^^^^^^^^^
  File "/var/home/sfc/Projects/azarrot-py312/src/azarrot/server.py", line 279, in create_server
    model_manager = ModelManager(config, backends)
                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/var/home/sfc/Projects/azarrot-py312/src/azarrot/models/model_manager.py", line 57, in __init__
    self.refresh_models()
  File "/var/home/sfc/Projects/azarrot-py312/src/azarrot/models/model_manager.py", line 204, in refresh_models
    model_info = backend.load_model(model)
                 ^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/var/home/sfc/Projects/azarrot-py312/src/azarrot/backends/transformers_based_backend.py", line 157, in load_model
    transformers_model: Any = model_class.from_pretrained(
                              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/var/home/sfc/Projects/azarrot-py312/.venv/lib/python3.12/site-packages/optimum/intel/openvino/modeling_base.py", line 485, in from_pretrained
    return super().from_pretrained(
           ^^^^^^^^^^^^^^^^^^^^^^^^
  File "/var/home/sfc/Projects/azarrot-py312/.venv/lib/python3.12/site-packages/optimum/modeling_base.py", line 438, in from_pretrained
    return from_pretrained_method(
           ^^^^^^^^^^^^^^^^^^^^^^^
  File "/var/home/sfc/Projects/azarrot-py312/.venv/lib/python3.12/site-packages/optimum/intel/openvino/modeling_decoder.py", line 319, in _from_transformers
    main_export(
  File "/var/home/sfc/Projects/azarrot-py312/.venv/lib/python3.12/site-packages/optimum/exporters/openvino/__main__.py", line 386, in main_export
    model = TasksManager.get_model_from_task(
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/var/home/sfc/Projects/azarrot-py312/.venv/lib/python3.12/site-packages/optimum/exporters/tasks.py", line 2283, in get_model_from_task
    model = model_class.from_pretrained(model_name_or_path, **kwargs)
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/var/home/sfc/Projects/azarrot-py312/.venv/lib/python3.12/site-packages/transformers/models/auto/auto_factory.py", line 564, in from_pretrained
    return model_class.from_pretrained(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/var/home/sfc/Projects/azarrot-py312/.venv/lib/python3.12/site-packages/transformers/modeling_utils.py", line 262, in _wrapper
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "/var/home/sfc/Projects/azarrot-py312/.venv/lib/python3.12/site-packages/transformers/modeling_utils.py", line 4400, in from_pretrained
    hf_quantizer.postprocess_model(model, config=config)
  File "/var/home/sfc/Projects/azarrot-py312/.venv/lib/python3.12/site-packages/transformers/quantizers/base.py", line 207, in postprocess_model
    return self._process_model_after_weight_loading(model, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/var/home/sfc/Projects/azarrot-py312/.venv/lib/python3.12/site-packages/transformers/quantizers/quantizer_gptq.py", line 111, in _process_model_after_weight_loading
    model = self.optimum_quantizer.post_init_model(model)
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/var/home/sfc/Projects/azarrot-py312/.venv/lib/python3.12/site-packages/optimum/exporters/openvino/__main__.py", line 345, in post_init_model
    from auto_gptq import exllama_set_max_input_length
ModuleNotFoundError: No module named 'auto_gptq'

I found that gptqmodel has the same method exllama_set_max_input_length as auto_gptq, so I just added an import trying gptqmodel first, and it works.

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you make sure to update the documentation with your changes?
• Did you write any new necessary tests?

[IlyasMoutawwakil]

thanks for the catch ! autogptq is being deprecated so we probably should also warn the user about that and advise to use gptqmodel instead.

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.