不支持嵌入模型的本地配置 #2262
Description
例行检查
- 我已确认目前没有类似 issue
- 我已确认我已升级到最新版本
- 我已完整查看过项目 README，尤其是常见问题部分
- [X ] 我理解并愿意跟进此 issue，协助测试和提供反馈
- [X ] 我理解并认可上述内容，并理解项目维护者精力有限，不遵循规则的 issue 可能会被无视或直接关闭
问题描述
不支持嵌入模型的本地配置 .
本地服务器192.168.146.101是用vllm部署的bge-m3的嵌入模型，能正常使用嵌入模型。参考如下验证
curl --location 'http://192.168.146.101:8000/v1/embeddings'
--header 'Content-Type: application/json'
--header 'Authorization: ••••••'
--data '{
"input": "The food was delicious The waiter was friendly",
"model": "bge-m3",
"encoding_format": "float"
}'
能正常返回如下：
{
"id": "embd-829fe107865d439dbad4e6248cce9d66",
"object": "list",
"created": 1745917648,
"model": "bge-m3",
"data": [
{
"index": 0,
"object": "embedding",
"embedding": [
-0.011810302734375,
0.023284912109375,
...
在oneapi系统(192.168.146.111:3000)里配置如下(注：有试过ollama，自定义渠道,proxy)
再次验证
curl --location 'http://192.168.146.111:3000/v1/embeddings'
--header 'Content-Type: application/json'
--header 'Authorization: ••••••'
--data '{
"input": "The food was delicious The waiter was friendly",
"model": "bge-m3",
"encoding_format": "float"
}'
返回报错
{
"error": {
"message": "bad response status code 404 (request id: 2025042917014089443205558410300)",
"type": "upstream_error",
"param": "404",
"code": "bad_response_status_code"
}
}