A tolerance-aware optimistic verification protocol makes floating-point neural-network inference verifiable on heterogeneous GPUs by accepting outputs within per-operator error bounds and resolving disputes via a Merkle-anchored dispute game.
Title resolution pending
1 Pith paper cite this work, alongside 36 external citations. Polarity classification is still indexing.
1
Pith paper citing it
36
external citations · OpenAlex
fields
cs.CR 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
TAO: Tolerance-Aware Optimistic Verification for Floating-Point Neural Networks
A tolerance-aware optimistic verification protocol makes floating-point neural-network inference verifiable on heterogeneous GPUs by accepting outputs within per-operator error bounds and resolving disputes via a Merkle-anchored dispute game.