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.

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

zkML: Trade-offs in accuracy vs. proving cost

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

Easy to use Verifiable AI and smart contracts interoperability.

To leverage AIGC technology with NFT, marketplace and privacy.

Curated list of ZKML models and use cases made with Giza

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

Reversion strategy for Haiko AMM

Benchmarking site for zkml frameworks

Optimization for on-chain private machine learning.