Config model path requires load from local

[wlm64]

The quick start examples don't seem to be working properly.

Loading test dataset from: dataset/nq/test.jsonl...
Traceback (most recent call last):
File "/home/jovyan/FlashRAG/examples/quick_start/simple_pipeline.py", line 40, in
pipeline = SequentialPipeline(config, prompt_template=prompt_templete)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/home/jovyan/FlashRAG/flashrag/pipeline/pipeline.py", line 53, in init
self.generator = get_generator(config)
^^^^^^^^^^^^^^^^^^^^^
File "/home/jovyan/FlashRAG/flashrag/utils/utils.py", line 48, in get_generator
with open(os.path.join(config["generator_model_path"], "config.json"), "r") as f:
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
FileNotFoundError: [Errno 2] No such file or directory: 'meta-llama/Meta-Llama-3-8B-Instruct/config.json'

It seems to only look for the model locally, but upon passing the direct model path there's a model import error later in the flow.

I was able to work around this by passing the config path directly in utils.py and using the hf path in the config but this seems janky. Is there a fix for this?

[ignorejjj]

Our approach is to first download the model locally (via hugginface cli), and then fill in the local model path into the YAML configuration file of Flashrag (or via config_dict).