Agree on the Model, Verify the Inference: GKR Protocols for HND-Based Transformer Inference

2026-07-23

AI Summary

This work addresses the challenges of model substitution and incomplete execution in outsourced Transformer inference by proposing the GKR-HND protocol, which uniquely integrates the GKR interactive proof system with homomorphic–non-homomorphic decomposition (HND). By introducing a model registration mechanism that binds weight commitments to client requests, the protocol enables a lightweight verifier to efficiently validate inference correctness under a non-collusion assumption, while delegating expensive public computations to designated nodes. This approach eliminates the need for dense matrix replay required by conventional methods, substantially reducing verification overhead without compromising inference completeness or model authenticity.

Abstract

Outsourced Transformer inference exposes clients to model substitution and incomplete execution, while direct replay removes the computational benefit of delegation. We present GKR-HND, a registered-model protocol for verifying the polynomial backbone of Homomorphic--Nonhomomorphic Decomposition Transformers. The retained verifier checks the GKR transcript and registered-weight openings, but delegates expensive public evaluations to an assigned computation worker. Assuming an honest retained verifier and prover--worker non-collusion, the verifier accepts only when the worker's signed, request-bound response agrees with the proof claims. Experiments with pretrained HND models validate the proof path and the delegated public computation without dense-matrix replay.

Problem

• model substitution
• incomplete execution
• outsourced inference
• verifiable computing
• Transformer

Innovation

• GKR protocols
• HND Transformers
• verifiable inference
• outsourced computation
• registered-model verification