fix passing hf_config args: #2547

[KrishnaM251]

Overview

This fix addresses issue #2547

I followed what Simon Mo suggested and modified EngineArgs and the ModelConfig class.

Here are the following changes i made:

args_utils.py

changed the EngineArgs class definition to include a new argument:
hf_kwargs: Optional[dict] = None
In create_engine_configs, I passed this in to the ModelConfig instantiation as follows:
hf_kwargs=(self.hf_kwargs or {})

vllm/config.py

In __ init__ function, I added the same argument:
hf_kwargs: Optional[dict] = None
I unpacked this in the get_config function call as follows:
**hf_kwargs

transformers_utils/config.py

added **hf_kwargs to the get_config function definition
inserted hf_kwargs into the AutoConfig.pretrained call

[simon-mo]

I don't think this does the job yet. We need the parameter to be passed in through LLM interface and the command line.

[hmellor]

@KrishnaM251 do you still plan to complete this work?

[simon-mo]

I believe so!

[github-actions]

This pull request has been automatically marked as stale because it has not had any activity within 90 days. It will be automatically closed if no further activity occurs within 30 days. Leave a comment if you feel this pull request should remain open. Thank you!

[mergify]

This pull request has merge conflicts that must be resolved before it can be
merged. @KrishnaM251 please rebase it. https://docs.github.com/en/pull-requests/collaborating-with-pull-requests/working-with-forks/syncing-a-fork