Vertex AI API C++ Client Library

An idiomatic C++ client library for the Vertex AI API, a service to train high-quality custom machine learning models with minimal machine learning expertise and effort.

While this library is GA, please note that the Google Cloud C++ client libraries do not follow Semantic Versioning.

Quickstart

The following shows the code that you'll run in the google/cloud/aiplatform/quickstart/ directory, which should give you a taste of the Vertex AI API C++ client library API.

#include "google/cloud/aiplatform/v1/endpoint_client.h"
#include <iostream>

int main(int argc, char* argv[]) try {
  if (argc != 3) {
    std::cerr << "Usage: " << argv[0] << " project-id location-id\n";
    return 1;
  }

  namespace aiplatform = ::google::cloud::aiplatform_v1;
  auto const location = std::string{argv[2]};
  auto client = aiplatform::EndpointServiceClient(
      aiplatform::MakeEndpointServiceConnection(location));

  auto const parent =
      std::string{"projects/"} + argv[1] + "/locations/" + location;
  for (auto r : client.ListEndpoints(parent)) {
    if (!r) throw std::move(r).status();
    std::cout << r->DebugString() << "\n";
  }

  return 0;
} catch (google::cloud::Status const& status) {
  std::cerr << "google::cloud::Status thrown: " << status << "\n";
  return 1;
}

Main classes

This library offers multiple *Client classes, which are listed below. Each one of these classes exposes all the RPCs for a gRPC service as member functions of the class. This library groups multiple gRPC services because they are part of the same product or are often used together. A typical example may be the administrative and data plane operations for a single product.

The library also has other classes that provide helpers, configuration parameters, and infrastructure to mock the *Client classes when testing your application.

Retry, Backoff, and Idempotency Policies.

The library automatically retries requests that fail with transient errors, and uses exponential backoff to backoff between retries. Application developers can override the default policies.

More Information