I’m tired of rolling my own infra to capture chat input and outputs, label it, and feed it to a training pipeline.
Is there a go to service / open source framework out there that will do all of this easier?
I’m tired of rolling my own infra to capture chat input and outputs, label it, and feed it to a training pipeline.
Is there a go to service / open source framework out there that will do all of this easier?
Not sure what kind of software you need, but here is a tool for ‘easy’ (by hand) gathering of chat datasets for later finetuning: oKatanaaa/lima-gui: A simple GUI utility for gathering LIMA-like chat data. (github.com)
Early work in progress, but may be worth a try.