部署本地嵌入模型jina-embeddings-v3报错400

[gzqy-hub]

System Info / 系統信息

docker+wsl2+ubuntu22.04
CUDA13.3
python3.12.13

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

最新版本3.3.1

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

Reproduction / 复现过程

配置上图参数后容器报错日志如下，“trust_remote_code=True”怎么设置？
2026-09-03T03:14:57.370Z ERROR xinference.api.restful_api pid:1 role:local address:0.0.0.0:9997 node:5fbe1d6d2bf3 [address=0.0.0.0:36367, pid=256] The model /data/xinference/modelscope/hub/models/jinaai/jina-embeddings-v3 references the module class 'custom_st.Transformer', which is not part of Sentence Transformers. Importing it executes third-party code. You can inspect the repository content at /data/xinference/modelscope/hub/models/jinaai/jina-embeddings-v3.

Please pass the argument trust_remote_code=True to allow custom code to be run.

Traceback (most recent call last):

File "/usr/local/lib/python3.12/dist-packages/xinference/api/restful_api.py", line 1229, in launch_model

model_uid = await (await self._get_supervisor_ref()).launch_builtin_model(

            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^  

File "/opt/inference/xinference/model/embedding/sentence_transformers/core.py", line 300, in load

self._model = SentenceTransformer(

^^^^^^^^^^^^^^^^^

File "/usr/local/lib/python3.12/dist-packages/sentence_transformers/util/decorators.py", line 46, in wrapper

return func(*args, **kwargs)

^^^^^^^^^^^^^^^^^

File "/usr/local/lib/python3.12/dist-packages/sentence_transformers/sentence_transformer/model.py", line 194, in init

super().__init__(

File "/usr/local/lib/python3.12/dist-packages/sentence_transformers/base/model.py", line 236, in init

modules, self.module_kwargs = self._load_modules(

^^^^^^^^^^^^^^^^^

File "/usr/local/lib/python3.12/dist-packages/sentence_transformers/base/model.py", line 1081, in _load_modules

return self._load_config_modules(model_name_or_path, **load_kwargs)

^^^^^^^^^^^^^^^^^

File "/usr/local/lib/python3.12/dist-packages/sentence_transformers/base/model.py", line 1213, in _load_config_modules

module_class: type[Module] = self._load_module_class_from_ref(

^^^^^^^^^^^^^^^^^

File "/usr/local/lib/python3.12/dist-packages/sentence_transformers/base/model.py", line 1466, in _load_module_class_from_ref

return import_module_class(

^^^^^^^^^^^^^^^^^

File "/usr/local/lib/python3.12/dist-packages/sentence_transformers/util/misc.py", line 228, in import_module_class

raise ValueError(

ValueError: [address=0.0.0.0:36367, pid=256] The model /data/xinference/modelscope/hub/models/jinaai/jina-embeddings-v3 references the module class 'custom_st.Transformer', which is not part of Sentence Transformers. Importing it executes third-party code. You can inspect the repository content at /data/xinference/modelscope/hub/models/jinaai/jina-embeddings-v3.

Please pass the argument trust_remote_code=True to allow custom code to be run.

2026-09-03T03:14:57.374Z INFO uvicorn.access pid:1 role:local address:0.0.0.0:9997 node:5fbe1d6d2bf3 172.17.0.1:46126 - "POST /v1/models HTTP/1.1" 400

Expected behavior / 期待表现

已经分别测试在环境变量中添加trust_remote_code=True和附加参数: sentence_transformers中添加trust_remote_code=True，两次启动部署报同样错误，请教正确的配置参数，谢谢！