适配qwen2.5-vl系列模型？

[Typhoona]

您好，感谢您杰出的工作。
我在使用python run.py -dataset mimic_ex -data_path /data/tanfeng/Medical-Graph-RAG/dataset -grained_chunk -ingraphmerge -construct_graph
运行时出现问题，由于没有openai的api key，我想使用本地或者其他非chatgpt系列的模型如qwen2.5-vl等，请问是否可以适配？

[nguyenhoangchu]

Sorry, I was just passing by and saw your comment. I'm also trying to adapt it to run on vLLM. In their paper, they used various models such as Llama, so theoretically the answer should be yes. However, I feels a bit strange because none of the modules seem to be designed for easy model swapping and I have to change each one separately, and I’m not sure how they managed it.
Note: the embedding model they use is embeeding text gpt. U can use this model or find another embedding model