Power-Softmax is a new HE-compatible attention variant that permits training and inference of billion-parameter polynomial LLMs with performance matching standard transformers.
HElayers: A tile tensors framework for large neural networks on encrypted data
2 Pith papers cite this work, alongside 43 external citations. Polarity classification is still indexing.
2
Pith papers citing it
43
external citations · OpenAlex
citation-role summary
background 1
citation-polarity summary
verdicts
UNVERDICTED 2roles
background 1polarities
background 1representative citing papers
AEGIS reduces inter-GPU communication by up to 81.3% in self-attention and reaches 96.62% scaling efficiency with 3.86x speedup on four GPUs for 2048-token encrypted Transformer inference.
citing papers explorer
-
Power-Softmax: Towards Secure LLM Inference over Encrypted Data
Power-Softmax is a new HE-compatible attention variant that permits training and inference of billion-parameter polynomial LLMs with performance matching standard transformers.
-
AEGIS: Scaling Long-Sequence Homomorphic Encrypted Transformer Inference via Hybrid Parallelism on Multi-GPU Systems
AEGIS reduces inter-GPU communication by up to 81.3% in self-attention and reaches 96.62% scaling efficiency with 3.86x speedup on four GPUs for 2048-token encrypted Transformer inference.