liuweikarlie / capstone-LLM

Record my learning for LLM

Readme

RAG folder explain

• embedding.py is for split the text and then convert it to embedding format, store it in vector database
• CustomLLM.py is load the LLM for langchain use
• RAG.py --> RAG framework (called the CustomLLM)
• Frontend.py ---> frontend webside to load the model and interact

Data folder

check the data folder readme

load_model folder

• you should download the llama.cpp
• get the gguf model
• check the load_model folder readme

Reading

• 数学计算+ReAct
• translate agent （Reflection）- Andrew Ng 做的，看了好像并不太难，用的chatgpt，但是听说会很慢