Papers · Proving training · VeriLoRA / zkLoRA

VeriLoRA / zkLoRA

VeriLoRA: Fine-Tuning Large Language Models with Verifiable Security via Zero-Knowledge Proofs

Guofu Liaounverified

Venue
NDSS 2026
Date
2025-08-29
Proof system
Lookup arguments + sum-check + polynomial commitments
Hardware
AMD EPYC 9654, NVIDIA A100 (80GB), 500 GB RAM
Numbers from
survey
Paper
https://arxiv.org/abs/2508.21393
PDF
https://arxiv.org/pdf/2508.21393

Reported benchmarks

modelparamsbatch sizeproving time sverification time s
LLaMA-2 (<=13B)13B16004

proving_time_note: <600s per sample

Notes

Claims the first end-to-end ZKP for LoRA fine-tuning: forward propagation, backward gradient propagation and LoRA weight updates. Numbers below are from survey Table IV, not read from the paper.

Our reading

Citation neighbourhood

Builds on
  • none recorded
Cited by, in this corpus
  • none recorded

Edges are a proxy: paper A's text mentions B anywhere (body or bibliography). See the full graph.

Discussed in