@jeffcliff@shitposter.world @sun@shitposter.world @yonle@fedinet.waltuh.cyou The local LLMs I have tried are pretty decent at writing test classes. I think that's probably one of the better usecases.
Tests are a lot of work, lot of boilerplate bullshit, and you don't let the LLM actually change the actual code. And it's very obvious when it mistake.
Timeline
Post
Remote status
Context
1
@SuperDicq @jeffcliff @yonle I'm going to experiment with local llm for writing code. I've got a decent gpu
Replies
0Fetching replies…