EZKL proof service

EZKL is an engine for doing inference for deep learning models and other computational graphs in a zk-snark (ZKML).

This service allows you to create proof requests for EZKL models. Using infernet-ml, you can also verify the correctness of the proof.

Infernet Configuration

The service can be configured as part of the overall Infernet configuration in config.json.

{
    "log_path": "infernet_node.log",
    //...... contents abbreviated
    "containers": [
        {
            "id": "ezkl_proof_service",
            "image": "ritualnetwork/ezkl_proof_service:latest",
            "external": true,
            "port": "3000",
            "allowed_delegate_addresses": [],
            "allowed_addresses": [],
            "allowed_ips": [],
            "command": "--bind=0.0.0.0:3000 --workers=2",
            "env": {
                "HF_TOKEN": "your-huggingface-token",
                "ARTIFACT_DIRECTORY": "~/.cache/ritual"
            }
        }
    ]
}

Environment Variables

HF_TOKEN

• Description: The HuggingFace API token to use for downloading models. No need to set this variable if your ezkl artifacts (model file, compiled circuit, etc.) are public. Use this if your artifacts are in a private Huggingface repo.
• Default: None
• Example: get one from Huggingface.

ARTIFACT_DIRECTORY

• Description: The local directory to store huggingface artifacts in.
• Default: None
• Example: "~/.cache/ritual"

Artifacts

Refer to the EZKL artifact generation documentation for a tutorial on how to create your own ezkl artifacts.

Usage

This service only supports off-chain requests for the time-being. On-chain request support will be added in the future.

Offchain (web2) Request

Please note: The examples below assume that you have an Infernet Node running locally on port 4000.

PythonCLIcURL

You can make a proof request like so:

SERVICE_NAME = "ezkl_proof_service"

client = NodeClient("http://127.0.0.1:4000")

# Define inputs
proof_req = EZKLGenerateProofRequest(
    repo_id=repo_id,
    witness_data=WitnessInputData.from_numpy(
      input_vector=numpy.array([1.0380048, 0.5586108, 1.1037828, 1.712096])
    ),
)

ezkl_request = InfernetInput(
    source=JobLocation.OFFCHAIN,
    destination=JobLocation.OFFCHAIN,
    data=proof_req.model_dump(),
)

# Define request
job_request = JobRequest(
    containers=[SERVICE_NAME],
    data=ezkl_request.model_dump()
)

# Request the job
job_id = await client.request_job(job_request)

# Fetch results
result = (await client.get_job_result_sync(job_id))["result"]

# Get the proof
proof = result.get("ezkl_proof")

# Get the output
output = result.get("output")

With this, you can verify the proof like so:

from infernet_ml.zk.ezkl.ezkl_utils import verify_proof_from_repo

r = await verify_proof_from_repo(result.get("ezkl_proof"), repo_id=repo_id)

# Note that the sync flag is optional and will wait for the job to complete.
# If you do not pass the sync flag, the job will be submitted and you will receive a job id, which you can use to get the result later.
infernet-client job -c ezkl_proof_service -i input.json --sync

where input.json looks like this:

{
    "repo_id": "huggingface/Ritual-Net/ezkl_linreg_10_features",
    "witness_data": {
        "input_data": {
            "dtype": 1,
            "shape": [
                1,
                10
            ],
            "values": [
                0.06637126207351685,
                0.09468276053667068,
                0.974238395690918,
                0.737287163734436,
                0.8200138211250305,
                0.4537636339664459,
                0.0689813494682312,
                0.46179142594337463,
                0.21672245860099792,
                0.807859480381012,
            ]
        },
        "output_data": null
    },
    "vk_address": null
}

curl -X POST http://127.0.0.1:4000/api/jobs \
    -H "Content-Type: application/json" \
    -d '{
        "containers": ["ezkl_proof_service"],
        "data": {
            "source": "OFFCHAIN",
            "destination": "OFFCHAIN",
            "data": {
                "repo_id": "huggingface/Ritual-Net/ezkl_linreg_10_features",
                "witness_data": {
                    "input_data": {
                        "dtype": 1,
                        "shape": [
                            1,
                            10
                        ],
                        "values": [
                            0.06637126207351685,
                            0.09468276053667068,
                            0.974238395690918,
                            0.737287163734436,
                            0.8200138211250305,
                            0.4537636339664459,
                            0.0689813494682312,
                            0.46179142594337463,
                            0.21672245860099792,
                            0.807859480381012,
                        ]
                    },
                    "output_data": null
                },
                "vk_address": null
            }
        }
    }'