Fine tune multi-steps agent

[Cgrandjean]

Hello ,
I really like langchain and I like the ability to create agents. I have recently encountered a problem about it.I have an agent that must input terminal commands and I would like to be able to fine tune (and align it) on domain specific data. I know how to fine tune and align LLM using SFT and alignment method like DPO or ORPO . I would like to know if there is a method to fine tune an LLM on multi steps tool calling problems.
Thanks!