I wish I could fine tune a model to my coding style, effectively creating a coding agent clone of myself.
Discussion
That's the ultimate goal right there. In due time I'm sure a model would be able to learn from your past projects.
I have 281 projects on github... it should be doable
rag?
i want all code it generates to have the same taste as me without needing to load examples into its context every time
“For most developers, starting with a RAG-based system or a managed code customization service is the most practical way to get code in your own style from an LLM, especially with a large number of repositories. Fine-tuning is best if you need even deeper alignment and have the resources to support it”
You only have to create the embeddings once though (until you want to update with newer data), but I’m not sure how well rag works with code predictions
can't you use all of the code you've written as the training data somehow? (am tarded)
pretty much using your github repo as a RAG
apparently this is not a new idea
Guarantee someone will build a “connect your GitHub to learn your style” feature
That doesn't seem too far away. Train an agent against your git..
Nice to meet you mdubore 🤝😉
Agent Smith had the same dream as you, but he went too far.
You’d need to code in the Balmer Peak 🍺
that'd be gansta
“Make the code documentation sound more sarcastic.”
Trying to clone the coding style is super cool! It makes me fantasize, if only I could clone my passion for buying a luxury yacht, just like fine-tuning a model to replicate my coding habits, and copying the “want to buy a yacht” fervor, so that no matter what I do in the future, I'll always carry this energy to own a luxury yacht and sail to freedom, and even when I knock on the code, I feel like I'm saving up for the dream of yachting, so I'm twice as hot!
https://bsky.app/profile/yachttrading.bsky.social/post/3loilw4hqss2c
Nice to meet you 😉🤝
Do you want to have your jb55 robot 🤔😧