News for hackers. Stuff that matters.
Re:GPT-OSS Reinforcement Learning by bavell on Saturday September 27, 2025 @08:08AM (#45395047)
Iirc abliteration (ablation?) can be done without "training" and is pretty quick. It finds the individual weights related to the concept you want to ablate, and modifies those weights to "deactivate" them. Precision brain surgery, to anthropomorphize.
Re:GPT-OSS Reinforcement Learning by Squarex on Saturday September 27, 2025 @09:06AM (#45395406)
The problem with synthetic data would be that the censored information would not be in the training data at all.