[Bug]

[xuwengen0909]

📦 部署环境

Docker

📦 部署模式

服务端模式(lobe-chat-database 镜像)

📌 软件版本

v1.68.7

💻 系统环境

Ubuntu

🌐 浏览器

Chrome

🐛 问题描述

在做文本上传后，做向量化操作报错，需要怎么配置，默认他使用了https://api.openai.com/v1/embeddings ，这个可以在哪里配置

[rewrite] /repos/kb_ezGV6zl1M1Q9 -> http://127.0.0.1:3210/zh-CN__0__light/repos/kb_ezGV6zl1M1Q9
[rewrite] /repos/kb_ezGV6zl1M1Q9 -> http://127.0.0.1:3210/zh-CN__0__light/repos/kb_ezGV6zl1M1Q9
[rewrite] /repos/kb_ezGV6zl1M1Q9 -> http://127.0.0.1:3210/zh-CN__0__light/repos/kb_ezGV6zl1M1Q9
[rewrite] /repos/kb_ezGV6zl1M1Q9 -> http://127.0.0.1:3210/zh-CN__0__light/repos/kb_ezGV6zl1M1Q9
run embedding task 1
run embedding task 2
run embedding task 3
[rewrite] /repos/kb_ezGV6zl1M1Q9 -> http://127.0.0.1:3210/zh-CN__0__light/repos/kb_ezGV6zl1M1Q9
[rewrite] /files -> http://127.0.0.1:3210/zh-CN__0__light/files
[rewrite] /discover -> http://127.0.0.1:3210/zh-CN__0__light/discover
[rewrite] /chat -> http://127.0.0.1:3210/zh-CN__0__light/chat
embeddingChunks error {
message: '{"endpoint":"https://api.openai.com/v1","error":{"message":"request to https://api.openai.com/v1/embeddings failed, reason: connect ECONNREFUSED 199.59.148.7:443","type":"system","errno":"ECONNREFUSED","code":"ECONNREFUSED"},"errorType":"ProviderBizError","provider":"openai"}',
name: 'EmbeddingError'
}

📷 复现步骤

No response

🚦 期望结果

No response

📝 补充信息

No response

[lobehubbot]

👀 @xuwengen0909

Thank you for raising an issue. We will investigate into the matter and get back to you as soon as possible.
Please make sure you have given us as much context as possible.
非常感谢您提交 issue。我们会尽快调查此事，并尽快回复您。 请确保您已经提供了尽可能多的背景信息。

[dosubot[bot]]

@xuwengen0909 你好！我是Dosu，我是一个可以帮助你解决错误、回答问题并成为贡献者的机器人。我在这里帮助你，等待人类维护者。

To configure the AI provider for embeddings in the lobe-chat application, you can specify a different embedding model by adding a line to your .env file. For example, to use a different model, you can add:

DEFAULT_FILES_CONFIG="embedding_model=openai/text-embedding-3-large"

This configuration allows you to change the default embedding model used by the application. If you are experiencing ECONNREFUSED errors, it might be due to network issues or incorrect API endpoint configurations. Ensure that your network settings allow connections to the specified API endpoint and that the endpoint URL is correctly configured in your application settings ^[1].

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Bug Report | Other}

[dosubot[bot]]

I found a similar discussion that might be relevant to your issue. The discussion titled "[Bug] 将.txt文件加入知识库进行分块向量化时，向量化失败" was resolved by correcting the format of the API proxy address ^[1].

For your specific issue, it seems like the vectorization process is failing due to a connection error (ECONNREFUSED) when trying to access the OpenAI API. You can configure the API proxy by setting the OPENAI_PROXY_URL environment variable. This allows you to override the default OpenAI API request base URL. Here is an example configuration in a Docker Compose file:

version: '3.8'

services:
  lobe-chat:
    image: lobehub/lobe-chat
    container_name: lobe-chat
    restart: always
    ports:
      - '3210:3210'
    environment:
      OPENAI_API_KEY: sk-xxxx
      OPENAI_PROXY_URL: https://api-proxy.com/v1
      ACCESS_CODE: lobe66

Make sure to replace https://api-proxy.com/v1 with the correct proxy URL you intend to use. After setting this up, you should also configure the OpenAI proxy address and key in the AI service provider settings within the application interface ^[2][3].

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Bug Report | Other}

[xuwengen0909]

我不需要使用openi的embedding模型，我需要指定自定义的AI服务商的embedding模型，或者说是我自己的ollama 部署的embedding模型。

[lobehubbot]

Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑‍🤝‍🧑👫🧑🏿‍🤝‍🧑🏻👩🏾‍🤝‍👨🏿👬🏿

---

I don't need to use openi's embedding model, I need to specify a custom AI service provider's embedding model, or my own ollama deployment embedding model.

[xuwengen0909]

@dosu
我不需要使用openi的embedding模型，我需要指定自定义的AI服务商的embedding模型，或者说是我自己的ollama 部署的embedding模型。

[lobehubbot]

Bot detected the issue body's language is not English, translate it automatically. 👯👭🏻🧑‍🤝‍🧑👫🧑🏿‍🤝‍🧑🏻👩🏾‍🤝‍👨🏿👬🏿

---

@dosu
I don't need to use openi's embedding model, I need to specify a custom AI service provider's embedding model, or my own ollama deployment embedding model.

[xuwengen0909]

@xuwengen0909 你好！我是Dosu，我是一个可以帮助你解决错误、回答问题并成为贡献者的机器人。我在这里帮助你，等待人类维护者。

要配置 lobe-chat 应用程序中的嵌入 AI 提供商，您可以通过在您的 .env 文件中添加一行来指定不同的嵌入模型。例如，要使用不同的模型，您可以添加：

DEFAULT_FILES_CONFIG="embedding_model=openai/text-embedding-3-large"
此配置允许您更改应用程序使用的默认嵌入模型。如果您遇到 ECONNREFUSED 错误，可能是由于网络问题或错误的 API 端点配置。请确保您的网络设置允许连接到指定的 API 端点，并且端点 URL 已在您的应用程序设置中正确配置 [1]。

继续与Dosu交谈，请提及。

[帮助 Dosu 学习！提供反馈：精彩回应 | 无关答案 | 错误来源 | 过于冗长 | 幻觉 | 错误报告 | 其他]

我不需要使用openi的embedding模型，我需要指定自定义的AI服务商的embedding模型，或者说是我自己的ollama 部署的embedding模型。

[dosubot[bot]]

To specify a custom AI service provider's embedding model or integrate your own Ollama deployment in Lobe Chat, you can follow these steps:

1. Configure Custom Providers: Use the web UI interface to add custom providers. Fill out the form with the necessary details such as Service ID, Service Name, Description, Logo URL, request format, and API key. Currently, Lobe Chat does not support configuring custom model providers through environment variables ^[1].

2. Integrate Ollama Deployment:

   • Install Ollama: Ensure Ollama is installed on your local machine. You can find installation instructions on the Ollama official site.
   • Run Ollama Locally: Start Ollama on your local machine, typically on port 11434.
   • Run LobeChat with Docker: Use the following Docker command to start LobeChat and connect it to your local Ollama service:

     docker run -d -p 3210:3210 -e OLLAMA_PROXY_URL=http://host.docker.internal:11434 lobehub/lobe-chat

   • Configure Non-Local Access (Optional): If needed, set OLLAMA_ORIGINS and OLLAMA_HOST environment variables to allow cross-origin requests and specify the host and port ^[2].

By following these steps, you can configure Lobe Chat to use a custom AI service provider or your own Ollama deployment for embeddings.

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Bug Report | Other}

[wxxsfxyzm]

用oneapi中转一下或许是个好方法