qwen2-vl 系列无法awq量化

[kratorado]

Describe the feature
Please describe the feature requested here(请在这里描述需求)
尝试对finetune后的qwen2-vl-7b-instruct进行awq量化，会报错。错误信息见下方

Paste any useful information
Paste any useful information, including papers, github links, etc.(请在这里描述其他有用的信息，比如相关的论文地址，github链接等)

  File "/data/admin/work/swift-training/swift/swift/cli/export.py", line 5, in <module>
    export_main()
  File "/data/admin/work/swift-training/swift/swift/llm/export/export.py", line 40, in export_main
    return SwiftExport(args).main()
  File "/data/admin/work/swift-training/swift/swift/llm/base.py", line 45, in main
    result = self.run()
  File "/data/admin/work/swift-training/swift/swift/llm/export/export.py", line 25, in run
    quantize_model(args)
  File "/data/admin/work/swift-training/swift/swift/llm/export/quant.py", line 197, in quantize_model
    QuantEngine(args).quantize()
  File "/data/admin/work/swift-training/swift/swift/llm/export/quant.py", line 24, in __init__
    pt_engine, self.template = prepare_pt_engine_template(args, **kwargs)
  File "/data/admin/work/swift-training/swift/swift/llm/infer/utils.py", line 160, in prepare_pt_engine_template
    pt_engine: PtEngine = SwiftInfer.get_infer_engine(args, infer_backend='pt', load_model=load_model, **kwargs)
  File "/data/admin/work/swift-training/swift/swift/llm/infer/infer.py", line 70, in get_infer_engine
    return infer_engine_cls(**kwargs)
  File "/data/admin/work/swift-training/swift/swift/llm/infer/infer_engine/pt_engine.py", line 58, in __init__
    self.model, self.processor = get_model_tokenizer(
  File "/data/admin/work/swift-training/swift/swift/llm/model/register.py", line 494, in get_model_tokenizer
    model, processor = get_function(model_dir, model_info, model_kwargs, load_model, **kwargs)
  File "/data/admin/work/swift-training/swift/swift/llm/model/model/qwen.py", line 499, in get_model_tokenizer_qwen2_vl
    patch_output_clone(model.model.embed_tokens)
  File "/home/admin/miniconda3/envs/swift/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1729, in __getattr__
    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
AttributeError: 'Qwen2VLForConditionalGeneration' object has no attribute 'embed_tokens'

Additional context
Add any other context or information here(其他信息可以写在这里)
涉及到的包版本：

swift==git+https://github.com/modelscope/swift.git@66484498f8f3fab31ab5274da1b189a6b59a4dd6#egg=ms_swift
autoawq==0.2.7.post2
autoawq_kernels==0.0.7
transformers==4.47.0

查看了transformers的相应的源码，Qwen2VLForConditionalGeneration确实是没有embed_tokens这个属性的，倒是在 self.model 下有一个，也就是 Qwen2VLModel 有这个属性，不知道嫁接过去能不能行。

另外，qwen2-vl系列是有官方awq量化的，那他们是怎么做的呢

[Jintao-Huang]

试试gptq呗 我也没跑通

[kratorado]

试试gptq呗 我也没跑通

行吧

[yfllllll]

@kratorado 请问你解决了吗

[kratorado]

@kratorado 请问你解决了吗

没有

[dhhcj1]

好像本身就不支持awq， gptq 没问题