embedding模型接口调用后名称发生变化

[sliontc]

System Info / 系統信息

Python 3.10
Ubuntu 22.04

Running Xinference with Docker? / 是否使用 Docker 运行 Xinfernece？

• docker / docker
• pip install / 通过 pip install 安装
• installation from source / 从源码安装

Version info / 版本信息

Release: v0.15.4

The command used to start Xinference / 用以启动 xinference 的命令

xinference-local -h 0.0.0.0

Reproduction / 复现过程

1.运行部署xinference，然后launch bge-m3 embedding

2.用curl测试：curl -X POST "http://dev.xxx.cn:9997/v1/embeddings" -H "accept: application/json" -H "Content-Type: application/json" -d "{"model":"bge-m3","input":"What is the capital of China?"}"
3.返回结果：

{"object":"list","model":"bge-m3-1-0","data":[{"index":0,"object":"embedding","embedding":[-0.031030265614390373,0.035563819110393524,-0.04539928585290909,-0.010311655700206757,0.006988677196204662,0.05363959074020386,-0.025254059582948685,-0.008242975920438766,-0.0012899866560474038,-0.016217537224292755,0.0019480991177260876,0.05430838093161583,-0.009749211370944977,0.02197396382689476,-0.0310926772654056...]

Expected behavior / 期待表现

接口返回的model名字应该bge-m3

[codingl2k1]

This is the model replica name. If we return the model name instead of the model replica name, we won't know which model replica serves this request. Perhaps we can extend a field to include {"model": "bge-m3", "model_replica": "bge-m3-1-0"}?

@qinxuye

[sliontc]

Yes, it's a good idea. Since some third-party may use the returned model info to continue work.

[qinxuye]

That's a solution, we can let the user know the model and exact replica.

[github-actions]

This issue is stale because it has been open for 7 days with no activity.