[Question]: Does it support API Key for Xinference

[0000sir]

Describe your problem

I deployed models locally with Xinference, both LLM and embedding models, with API key authorization enabled.
When I configure RAGFlow for Xinference models, there's a 401 error occurred, I noticed a place holder in the text box says "for locally deployed model, ignore this", is that means there's no authorization support for local models?

The API key works fine when I configure it in OneAPI service.

Thank you for your help.

BTW: I found a QA says streaming output is not supported in RAGFlow and still working on it, when will we see it

[0000sir]

chat_model.py line 100

class XinferenceChat(Base):
def init(self, key=None, model_name="", base_url=""):
if not base_url:
raise ValueError("Local llm url cannot be None")
if base_url.split("/")[-1] != "v1":
base_url = os.path.join(base_url, "v1")
key = "xxx"
super().init(key, model_name, base_url)


It looks like the api key not used while sending requests.
Why you set key to xxx here??

[KevinHuSh]

Streaming is supported.