ezkl is an engine for doing inference for deep learning models and other computational graphs in a zk-snark (ZKML). Use it from Python, Javascript, or the command line.

Curated list of ZKML models and use cases made with Giza

Easy to use Verifiable AI and smart contracts interoperability.

Optimization for on-chain private machine learning.

Reversion strategy for Haiko AMM

Giza agents run provable ML models and act autonomously on Starknet using native Account Abstraction

To leverage AIGC technology with NFT, marketplace and privacy.

Securing confidential data in database using ZKML based cryptographic approach with auto encoder based encoding

zkML: Trade-offs in accuracy vs. proving cost

Benchmarking site for zkml frameworks

Code for collecting ezkl performance metrics on Zero-Knowledge Proof generation over inference tasks for multiple deep neural network models.