ATLAS: Automated Approximation of Transformers for Efficient Homomorphic Inference in One Hour

본문 미리보기

arXiv:2607.23478v1 Announce Type: new Abstract: Fully homomorphic encryption (FHE) provides strong cryptographic guarantees for private inference, but deploying transformer models under FHE remains prohibitively expensive. A key bottleneck is that non-linear operations such as softmax, normalization, and activation must be replaced with polynomial approximations compatible with the CKKS scheme, and the multiplicative depth consumed by these approximations dominates inference cost. Recent framew