k8s-specific-knowledge-base

This is a Question-Answering system based on k8s-specific knowledge build on ChatGLM2-6B, serving by Ray. It's not ready for production, but we'll target for this.

Architecture

🤔 What we used ?

• LangChain to build LLM applications.
• Ray for accelerating and serving.
• ChatGLM2-6B as a base model.
• BAAI/bge-base-zh-v1.5 for embedding in semantic search.
• FAISS as a vector database.

📦 Corpus Including

• Kubernetes Website
• Kubernetes Blogs
• Kubernetes Books (Only for research usage)

🔖 What's next ?

• Containerization, FAISS is for single node.
• More efficient text splitting ways designed for Chinese.
• More approaches to support semantic search, e.g. key-word embeddings.
• Raw data management, like new uploading and deleting.
• Vector data persistent.
• Continuous Pre-Training.
• and so on ...

📚 Reference

• langchain-ray