Meta Learning for complicated problems

[flowermlh]

How to make LLM able to learn the prompt itself?
Langchain is good at telling LLM a prompt or prompt chains to act 。
When the problem is very complicated, human have to give each instructions to LLM，which is difficult and tedious。