Secretvoice1.0: Non-interactive secure transformer inference¶
Authors: W. Yang; Z. Liu; Y. Zhang; X. Bie; D. Yao
Year: 2026
status: auto
Links: link
System card: SecretVoice 1.0¶
| Group | 0. CKKS LLM frameworks surveyed in the SoK |
| Family | CKKS · Model-modifying (N-MOD) |
| Arithmetic | Approximate real (CKKS) |
| Interaction | non-interactive FHE |
| Model scale | Original Transformer encoder-decoder |
| Library | Microsoft SEAL |
| Hardware | 1-core CPU |
| Packing | L-ROW |
| Latency | 1488 s per encoder block (32 tokens) |
| Open source | no |
| Status | Niche |
| Relation to the SoK | Fuses linear layers into StC-first bootstrapping. |
| Note | Operation fusion trades memory for latency; no utility metric. |
| Systems page | SecretVoice 1.0 |
SoK reference [69]¶
W. Yang, Z. Liu, Y. Zhang, X. Bie, and D. Yao, “Secretvoice1.0: Non-interactive secure transformer inference,” TechRxiv, 2026.
Listed on the reference page of the site.
Files¶
No local PDF in the archive.
Database key: secretvoice1-0-non-interactive-secure-transformer-inference · created 2026-11-05 · updated 2026-11-05 · edit with python3 tools/paperdb.py set secretvoice1-0-non-interactive-secure-transformer-inference COLUMN VALUE