Deprecate vocab_size in other two VLMs

[zucchini-nlp]

What does this PR do?

Continuing on #31655, let's deprecate the model.config.vocab_size for all VLMs for consistency across repo and use model.config.text_config.vocab_size in future models.

Currently only Fuyu and Paligemma have vocab_size in config file, and I depreacted it to v4.43. Or should it be v4.44?

[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.

[amyeroberts]

Very nice - thanks for handling!

[prakash-shubham-2003]

@zucchini-nlp @amyeroberts
After this update, I am facing an error while making an instance of LlavaProcessor. This is the code I wrote:
'processor = AutoProcessor.from_pretrained("llava-hf/llava-1.5-7b-hf")'

It gives the following error:
'TypeError: LlavaProcessor.init() got an unexpected keyword argument 'chat_template''

It was working fine before the update. I have the following versions of transformers and bitsandbytes installed:
'!pip install -q -U transformers==4.37.2'
'!pip install -q bitsandbytes==0.41.3 accelerate==0.25.0'

Is there any issue with the current update or is there any other solution for this problem?

[zucchini-nlp]

Answered the question in another thread, not related to this PR. Please open an issue if you still encounter the error after updating transformers :)