LATIDIA · Ciberseguridad

Una implementación de FHE de código abierto de extremo a extremo para la inferencia de llama 3 8B que preserva la privacidad

arXiv:2609.12378v1 Anuncio Tipo: nuevo Resumen: Los servicios LLM en la nube suelen requerir que los usuarios envíen indicaciones a un proveedor de modelos, lo que crea un riesgo para la privacidad. El cifrado totalmente homomórfico (FHE) permite que un servidor realice inferencias.

La noticia

arXiv:2609.12378v1 Announce Type: new Abstract: Cloud LLM services typically require users to send prompts to a model provider, creating a privacy risk. Fully homomorphic encryption (FHE) lets a server perform inference without decrypting the input, but representing data as ciphertexts adds storage and computational overhead. In CKKS-based LLM inference, the packing scheme maps logical tensors to ciphertexts and slots. It therefore determines the ciphertext count and the homomorphic cost of linear layers, and it constrains how data pass between linear layers, attention, and nonlinear computation. As models and sequences grow, inefficient layouts accumulate encoding, compute, and layout-conversion overhead. We present Odin, an FHE inference system that co-designs ciphertext packing and