jlcbj

Milvus 2.4 support hybrid search

https://milvus.io/docs/embed-with-bm25.md

# parser.add_argument("--device", type=str, default=None, help="device") comment it in vllm_adapter.py will fix the issue

> > # parser.add_argument("--device", type=str, default=None, help="device") > > comment it in vllm_adapter.py will fix the issue > > vllm_adapter.py 该文件包含parser.add_argument("--device", type=str, default=None, help="device")该语句，不知道你表达的是什么 字面意思不是很清楚，device参数重复了。 前面 parser = AsyncEngineArgs.add_cli_args(parser) 看vllm包的参数定义：...

> The problem of parser has been resolved and the vllm can be loaded. > > However, new issue occurs: > > LLMServer Generate Error, Please CheckErrorInfo.: 'TokenizerGroup' object has...

model codellama-13b-sql-sft has the problem,can not format output according to the instruction. change another model should work