Halo2 Deep Learning

This is a proof-of-concept implementation of inference for deep learning models in a zk-snark using Halo2. 2d convolution, fully connected (affine) layers, and nonlinearities such as ReLU and sigmoid are implemented. The input image and model parameters are provided as private advice and the last layer is the public input (instance column). Other configurations are also possible.

We give an example of proving inference with a model that achieves 97.5% accuracy on MNIST in the examples.

Note that the library requires a nightly version of the rust toolchain. You can change the default toolchain by running:

rustup override set nightly         

Running examples

The MNIST inference example is contained in examples/conv2d_mnist. To run it:

cargo run --release --example conv2d_mnist

We also provide an example which runs an MLP on input data with four dimensions. To run it:

cargo run --release --example mlp_4d

Running onnx example

To run the example which loads parameters from an ONNX model file you need to enable the onnx build feature:

cargo run --release --example smallonnx --features onnx

Running benchmarks

We include proof generation time benchmarks for some of the implemented layers including the affine, convolutional, and ReLu operations (more to come).

To run these benchmarks:

cargo bench

To run a specific benchmark append one of affine, cnvrl, relu to the command. You can then find benchmarks results and plots in target/criterion. Note that depending on the capabilities of your machine you may need to increase the target time on the Criterion config. For instance:

criterion_group! {
  name = benches;
  config = Criterion::default().measurement_time(Duration::from_secs(10));
  targets = runrelu
}

Docs

Use cargo doc --open --feature onnx to compile and open the docs in your default browser.