Using openevolve + unsloth for training

[alientony]

Would it be possible that we can incorporate using unsloth as a lora trainer addition?

With the island system, complex prompt creation through previous code examples, and proper safe guards to prevent the llms from gaming the system.

Doesn't this provide a a great opportunity for a training loop and with the already provided ground work laid out. Just need to build a bridge.

I'm working with a notebook that is taking advantage of the circle packing
Just now, I'm using the evaluator not the entire system. Still working on that.