VCD in LVLMs

[sunzjz]

请问VCD要用在其他多模态模型上的话，这个第2步是要修改哪些，写的好像不是很清楚。
2. Slightly modify llava_llama.py:
a. Add contrastive decoding parameters in the LlavaLlamaForCausalLM class's forward function to avoid exceptions in model.generate.
b. Add the prepare_inputs_for_generation_cd function.

略过这一步的话，运行报错“ValueError: The following model_kwargs are not used by the model: ['images_cd', 'cd_alpha', 'cd_beta'] (note: typos in the generate arguments will also show up in this list)”。这些参数貌似没有在其他LVLM里使用？