Add Batched Inference Script for VCoder

[dg845]

This PR adds a script to perform batched inference from a VCoder checkpoint in vcoder_llava/eval/, based on vcoder_llava/eval/model_seg_loader.py and vcoder_llava/eval/model_depth_loader.py.

[pribadihcr]

Hi, I run the following example script:

python model_batched_inference.py --model-path shi-labs/vcoder_llava-v1.5-7b --load-4bit --image-folder ../input_folder --output-folder ./out_vcoder --prompt "describe this image in details" 

I got the following error:

File "/VCoder/model_batched_inference.py", line 238, in evaluate_model
output_ids = model.generate(
File "/Anaconda3/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
return func(*args, **kwargs)
File "/Anaconda3/lib/python3.10/site-packages/transformers/generation/utils.py", line 1282, in generate
self._validate_model_kwargs(model_kwargs.copy())
File "/Anaconda3/lib/python3.10/site-packages/transformers/generation/utils.py", line 1155, in _validate_model_kwargs
raise ValueError(
ValueError: The following model_kwargs are not used by the model: ['depths'] (note: typos in the generate arguments will also show up in this list)