Afonso Domingues
Titles as the sources gave them. In 1 entry, with 1 checked quotation.
What they said
Newest first. Each quotation was found word for word in the source.
“we find that models with similar broad misalignment rates can exhibit very different behavioural profiles that a single score cannot capture.”
In the Base
Every entry they appear in.
Not All Misalignment Is Alike: A Decomposed Benchmark for Emergent Misalignment
According to the abstract, Domingues and five co-authors introduce EMERGE, a benchmark that tracks eight separate misaligned behaviours when models are fine-tuned on narrow tasks, and apply it to 8 open-weight models and 13 fine-tuning datasets.