nexent ,bug in the local knowledge base tool #1548
-
|
I perform local deployment of Nexent via Docker in the Ubuntu system. |
Beta Was this translation helpful? Give feedback.
Replies: 3 comments 4 replies
-
Beta Was this translation helpful? Give feedback.
-
Beta Was this translation helpful? Give feedback.
-
|
Hi there, could you share the document you use to create knowledgebase with us? The links above are not working. @ganzhipeng111 |
Beta Was this translation helpful? Give feedback.
-
Beta Was this translation helpful? Give feedback.
-
|
Word documents can be uploaded and imported into the library, while PDF documents cannot. |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for replying. This problem can be simply avoided by using embedding model from Jina series (Recommended jina-embedding-v3). You can apply for 1,000,000 embedding tokens from https://jina.ai/zh-CN/ for free, or use our one-click trial website https://try.nexent.tech/zh. |
Beta Was this translation helpful? Give feedback.
-
|
sorry,Jina-embedding-v3 is not free, and it is very expensive.It’s a bit unaffordable for me, and the models from Guiji Liudong are more suitable for me.So, how to solve this problem? Using other embedding models can also enable the import of PDF documents into the library. |
Beta Was this translation helpful? Give feedback.
Thanks for replying. This problem can be simply avoided by using embedding model from Jina series (Recommended jina-embedding-v3).
More specifically, it is caused by the embedding model "text-embedding-v4", which only supports 25 parallel text messages in a single request. Nexent, on the other hand, has a default value 64 for that, causing an embedding error.
You can apply for 1,000,000 embedding tokens from https://jina.ai/zh-CN/ for free, or use our one-click trial website https://try.nexent.tech/zh.