[vLLM] metadata script

[mishig25]

Follow up to #957. To precisely show vLLM snippet on HF models, we need to have record on which models are supported by vLLM. vLLM provides supported models list: doc page, python src.

This PR creates a github cron job (runs once a day) script that:

1. parses python src through AST to extract supported architectures list (both for transformers & ggml/llama.cpp styles)
2. uploads the result to huggingface/vllm-metadata. You can see working example here.

3. HF models will use this list to whether show or not show vLLM snippet.

Find the python script here: get_vlmm_metadata.py.zip (you can locally test run if you want).

[mishig25]

@simon-mo could you confirm that this list is the correct way to find vLLM supported models both for transformers & ggml/llama.cpp architectures

_MODELS = {
    **_TEXT_GENERATION_MODELS,
    **_EMBEDDING_MODELS,
    **_MULTIMODAL_MODELS,
    **_SPECULATIVE_DECODING_MODELS,
}

[simon-mo]

Yes!

[Vaibhavs10]

Looks dope! I didn't test the script locally, but went through the exported dataset - looks amazing!

[pcuenca]

Suggested change

	architectures = [item for tup in models_dict.values() for item in tup]
	architectures = sorted(list({item for tup in models_dict.values() for item in tup}))

Maybe, if we want to remove duplicates and assuming tuple order does not matter (i.e., llama does not have to appear before LlamaForCausalLM

[osanseviero]

Very cool 🔥

[osanseviero]

Can we move this code to a script rather than having the Python code in the yaml? It will be easier to maintain, update, and review

[mishig25]

it will be easier to review

agree with this point

it will be easier to maintain, update

I think maintaining a separate python script would be painful. We would need to find a place to place this python script and tell the yaml job to download and run this python script (which can introduce other security issues since we are running what's being downloaded)

[osanseviero]

Suggested change

	class MODELS_SUB_LIST_VISITOR(ast.NodeVisitor):
	class ModelsSubListVisitor(ast.NodeVisitor):