Open Distillation of Hereditary Traits

·Alignment Forum··

TL;DRJosh and Neel show that distillation from a teacher model to a base pretrained student model transfers some of the teacher model’s traits (such as displaying negative emotion in the Gemma Needs Help evals)On its own this is pretty unsurprising, but Josh and Neel additionally show that even filtering out all the prompts and rollouts where the trait is mentioned doesn’t generally prevent the trait transferIn this post, I show a simple way to replicate and study these phenomena without access ...

Read full article →

Related Articles

How could the US and China verifiably slow AI?
josefely · LessWrong · 37m ago
Red-Teaming OpenAI's Internal Agent Monitoring Systems
Buck Shlegeris · Redwood Research · 6h ago
Monitorability evaluations and policy proposals
evhub · Alignment Forum · 8h ago
MIT's New Method Flags AI Models Trained on CASM Without Generating It
sdoering · Hacker News · 3mo ago
Neural nets are exceptionally vulnerable to goal self-hack
Stuart_Armstrong · Alignment Forum · 1d ago