Durable Quantization Conditioned Misalignment Attack on Large Language Models

Publication
The 13th International Conference on Learning Representations (ICLR)