[Question]: Fail to access embedding model(bge-large-zh-v1.5).Connection error.

[qifuxiao]

Describe your problem

我使用docker compose顺利完成了ragflow的的部署，并且使用ollama完成了添加模型，测试任务正常，我现在想使用Xinference框架进行模型管理，当我添加的时候报错Fail to access embedding model(bge-large-zh-v1.5).Connection error.

我查看了ragflow-server的日志：
NoneType: None
2025-02-28 10:04:43,961 INFO 14 172.19.0.6 - - [28/Feb/2025 10:04:43] "POST /v1/llm/add_llm HTTP/1.1" 200 -
2025-02-28 10:07:08,366 INFO 14 172.19.0.6 - - [28/Feb/2025 10:07:08] "GET /v1/user/info HTTP/1.1" 200 -
2025-02-28 10:07:08,374 INFO 14 172.19.0.6 - - [28/Feb/2025 10:07:08] "GET /v1/llm/my_llms HTTP/1.1" 200 -
2025-02-28 10:07:08,375 INFO 14 172.19.0.6 - - [28/Feb/2025 10:07:08] "GET /v1/user/tenant_info HTTP/1.1" 200 -
2025-02-28 10:07:08,397 INFO 14 172.19.0.6 - - [28/Feb/2025 10:07:08] "GET /v1/llm/factories HTTP/1.1" 200 -
2025-02-28 10:07:18,369 INFO 14 Retrying request to /embeddings in 0.867682 seconds
2025-02-28 10:07:19,239 INFO 14 Retrying request to /embeddings in 1.911701 seconds
2025-02-28 10:07:21,154 ERROR 14
Fail to access embedding model(bge-large-zh-v1.5).Connection error.
NoneType: None
2025-02-28 10:07:21,156 INFO 14 172.19.0.6 - - [28/Feb/2025 10:07:21] "POST /v1/llm/add_llm HTTP/1.1" 200 -
2025-02-28 10:08:03,862 INFO 14 172.19.0.6 - - [28/Feb/2025 10:08:03] "GET /v1/user/info HTTP/1.1" 200 -
2025-02-28 10:08:03,873 INFO 14 172.19.0.6 - - [28/Feb/2025 10:08:03] "GET /v1/llm/my_llms HTTP/1.1" 200 -
2025-02-28 10:08:03,874 INFO 14 172.19.0.6 - - [28/Feb/2025 10:08:03] "GET /v1/user/tenant_info HTTP/1.1" 200 -
2025-02-28 10:08:03,898 INFO 14 172.19.0.6 - - [28/Feb/2025 10:08:03] "GET /v1/llm/factories HTTP/1.1" 200 -
2025-02-28 10:08:49,277 INFO 14 172.19.0.6 - - [28/Feb/2025 10:08:49] "GET /v1/user/info HTTP/1.1" 200 -
2025-02-28 10:08:49,287 INFO 14 172.19.0.6 - - [28/Feb/2025 10:08:49] "GET /v1/user/tenant_info HTTP/1.1" 200 -
2025-02-28 10:08:49,289 INFO 14 172.19.0.6 - - [28/Feb/2025 10:08:49] "GET /v1/llm/my_llms HTTP/1.1" 200 -
2025-02-28 10:08:49,304 INFO 14 172.19.0.6 - - [28/Feb/2025 10:08:49] "GET /v1/llm/factories HTTP/1.1" 200 -
2025-02-28 10:08:51,718 INFO 14 Retrying request to /embeddings in 0.913683 seconds
2025-02-28 10:08:52,634 INFO 14 Retrying request to /embeddings in 1.858031 seconds
2025-02-28 10:08:54,495 ERROR 14
Fail to access embedding model(bge-large-zh-v1.5).Connection error.
这里并没有给出有用的报错信息

[KevinHuSh]

Give it a try: 172.17.0.1:9997/v1

[syks0121]

Give it a try: 172.17.0.1:9997/v1

I have a similar situation as well. How can I resolve it? It's also about Docker deployment.

ragflow-server log：
2025-03-01 00:48:19 2025-03-01 00:48:19,573 INFO 29 Retrying request to /embeddings in 0.773104 seconds
2025-03-01 00:48:20 2025-03-01 00:48:20,349 INFO 29 Retrying request to /embeddings in 1.601792 seconds
2025-03-01 00:48:21 2025-03-01 00:48:21,955 ERROR 29
2025-03-01 00:48:21 Fail to access embedding model(bge-large-zh-v1.5).Connection error.
2025-03-01 00:48:21 NoneType: None
2025-03-01 00:48:21 2025-03-01 00:48:21,957 INFO 29 172.18.0.6 - - [01/Mar/2025 00:48:21] "POST /v1/llm/add_llm HTTP/1.1" 200 -

[iopenet]

Windows10, Doker, Local Models.

1. Follow steps in docs to install ragflow-slim

2. Run local models with the desktop app of your choice

   • your model
   • your text-embedder
   • etc.

3. Run RagFlow

   • open http://127.0.0.1/
   • Sign up login : create new user

4. RagFlow admin

   • go to /Model providers

5. Select Models to be added

6. Added Models > click Add LLM (repeat for chat, embedding, etc)

   • model type : chat
   • model name : usually provided by your desktop app
   • base url : provided by your desktop app e.g. http://127.0.0.1:[port number]
   • max tokens

7. Click button System Model Settings

   • select the models added in step 6

Edit
Fixed with embedder update :
https://huggingface.co/kalle07/embedder_collection

hint : 102 is everywhere but not useful !
Instruction is misleading, no link to option on the GUI.
Please add both embedding model and LLM in Settings > Model providers firstly. Then, set them in 'System model settings'.