I would like to use Kernel Memory to build a chatbot for RAG, is it possible to configure Kernel Memory to store the chat history?

[chingcuimn]

I have created a asp.net project using Kernel Memory to import PDF documents and store the vectors in a file storage. When creating a chatbot to query the data from these PDF documents, I found out the chat is not storing the history. Is there a way that we can do that using kernel memory? The reason I don't want to go to the semantic kernel plugin route is the memory plugin is not working well for ollama (LLAMA LLM). Can someone help?

[marcominerva]

Can you take a look at https://github.com/marcominerva/KernelMemoryService. In this project, I'm using a MemoryCache to save conversation history: https://github.com/marcominerva/KernelMemoryService/blob/master/KernelMemoryService/Services/ChatService.cs

[chingcuimn]

Thanks so much for the reference. I modified my code and it worked well now using Azure OpenAI. Do you have some examples to use Ollama (Llama) as the text generator? I was trying to use Ollama connector for Kernel Memory, the results I got are not as good as using Azure OpenAI.

[marcominerva]

Do you mean this https://github.com/microsoft/kernel-memory/blob/main/extensions/Ollama/README.md? I don't have direct experience with it, I have always used Azure OpenAI.