Paper

Spectral Unlearning: Transformer Structure-Preserving Updates for Language Models · ByteTrail