deepseek Deepseek Knowledge Base Integration Configuration Method

[baike]Apply for DeepSeek API[/baike]
First, you need to apply for an API key on the DS official website. The current cost is the lowest among similar large models, equivalent to 1/50 of OpenAI O1.
Application address: https://platform.deepseek.com/

[baike]Set up a chat client[/baike]
After obtaining the API, you need to find a client of the runtime framework type to run the API. There are many on GitHub; I recommend one I often use, cherry-studio. After downloading it, enter the API you applied for from DS, and now you can use it in the local client.
Download address: https://cherry-ai.com/

[baike]Configure the embedding model[/baike]
Next, add an embedding model. The role of the embedding model is to vectorize and retrieve our local knowledge documents; you can understand it as data processing before DS is used.
We can directly select SiliconFlow in cherry-studio. You also need to create an API key, fill it back in after creating it, and then enter BBA/bge-m3 in the model addition field below.


[baike]Add knowledge base documents[/baike]
Next, you can add knowledge documents.