部署模型

初始模型部署

如果您想要将现有模型更新为最新预测服务（2019 年 9 月发布），则可以取消部署该模型并重新部署。

自正式版发布之日起，自定义模型的最长使用期限为 18 个月。该时间期限过后，您必须创建并训练新模型，才能继续对内容进行分类。

节点数将受“配额”页面上列出的配额限制；默认情况下，一个用户最多可以获得 10 个节点。

创建（训练）模型后，您必须先部署模型，然后才能对其进行在线（或同步）调用。

现在，如果您还需要其他在线预测能力，还可以更新模型部署。

网页界面

1. 导航至标题栏下方的测试和使用标签页。
2. 选择部署模型按钮。随即将打开一个新部署选项窗口。
3. 在新打开的部署选项窗口中，指定要在其中部署模型的节点数。 每个节点每秒可支持一定数量的预测查询 (QPS)。

   对于大多数实验性数据流量来说，一个节点时通常能够满足需要。

   

4. 选择部署以开始部署模型。

   
5. 模型部署操作完成后，您会收到电子邮件通知。

REST 和命令行

在使用下面的任何请求数据之前，请先进行以下替换：

• project-id：您的 GCP 项目 ID。
• model-id：您的模型的 ID（从创建模型时返回的响应中获取）。此 ID 是模型名称的最后一个元素。 例如：
  • 模型名称：projects/project-id/locations/location-id/models/IOD4412217016962778756
  • 模型 ID：IOD4412217016962778756

字段注意事项：

• nodeCount - 需要在其上部署模型的节点的数量。值必须介于 1 到 100 之间，包括 1 和 100。节点是机器资源的抽象，可处理模型的 qps_per_node 中指定的每秒在线预测查询次数 (QPS)。

注意：节点数量将受配额页面上列出的配额的限制；默认情况下，用户最多可以获得 10 个节点。

HTTP 方法和网址：

POST https://automl.googleapis.com/v1/projects/project-id/locations/us-central1/models/model-id:deploy

请求 JSON 正文：

{
  "imageClassificationModelDeploymentMetadata": {
    "nodeCount": 2
  }
}

如需发送请求，请选择以下方式之一：

curl

注意：请确保已将 GOOGLE_APPLICATION_CREDENTIALS 环境变量设置为您的服务帐号私钥文件路径。

将请求正文保存在名为 request.json 的文件中，然后执行以下命令：

curl -X POST \
-H "Authorization: Bearer "$(gcloud auth application-default print-access-token) \
-H "Content-Type: application/json; charset=utf-8" \
-d @request.json \
https://automl.googleapis.com/v1/projects/project-id/locations/us-central1/models/model-id:deploy

PowerShell

注意：请确保已将 GOOGLE_APPLICATION_CREDENTIALS 环境变量设置为您的服务帐号私钥文件路径。

将请求正文保存在名为 request.json 的文件中，然后执行以下命令：

$cred = gcloud auth application-default print-access-token
$headers = @{ "Authorization" = "Bearer $cred" }

Invoke-WebRequest `
  -Method POST `
  -Headers $headers `
  -ContentType: "application/json; charset=utf-8" `
  -InFile request.json `
  -Uri "https://automl.googleapis.com/v1/projects/project-id/locations/us-central1/models/model-id:deploy" | Select-Object -Expand Content

您应该会看到类似如下所示的输出。可以使用操作 ID 来获取任务的状态。如需查看示例，请参阅处理长时间运行的操作。

{
  "name": "projects/project-id/locations/us-central1/operations/operation-id",
  "metadata": {
    "@type": "type.googleapis.com/google.cloud.automl.v1.OperationMetadata",
    "createTime": "2019-08-07T22:00:20.692109Z",
    "updateTime": "2019-08-07T22:00:20.692109Z",
    "deployModelDetails": {}
  }
}

您可以通过以下 HTTP 方法和网址获取操作的状态：

GET https://automl.googleapis.com/v1/projects/project-id/locations/us-central1/operations/operation-id

操作完成的状态将类似于以下内容：

{
  "name": "projects/project-id/locations/us-central1/operations/operation-id",
  "metadata": {
    "@type": "type.googleapis.com/google.cloud.automl.v1.OperationMetadata",
    "createTime": "2019-06-21T16:47:21.704674Z",
    "updateTime": "2019-06-21T17:01:00.802505Z",
    "deployModelDetails": {}
  },
  "done": true,
  "response": {
    "@type": "type.googleapis.com/google.protobuf.Empty"
  }
}

Go

在试用此示例之前，请按照客户端库页面中与此编程语言对应的设置说明执行操作。

在 GitHub 上查看 反馈

import (
	"context"
	"fmt"
	"io"

	automl "cloud.google.com/go/automl/apiv1"
	automlpb "google.golang.org/genproto/googleapis/cloud/automl/v1"
)

// deployModel deploys a model.
func deployModel(w io.Writer, projectID string, location string, modelID string) error {
	// projectID := "my-project-id"
	// location := "us-central1"
	// modelID := "TRL123456789..."

	ctx := context.Background()
	client, err := automl.NewClient(ctx)
	if err != nil {
		return fmt.Errorf("NewClient: %v", err)
	}
	defer client.Close()

	req := &automlpb.DeployModelRequest{
		Name: fmt.Sprintf("projects/%s/locations/%s/models/%s", projectID, location, modelID),
	}

	op, err := client.DeployModel(ctx, req)
	if err != nil {
		return fmt.Errorf("DeployModel: %v", err)
	}
	fmt.Fprintf(w, "Processing operation name: %q\n", op.Name())

	if err := op.Wait(ctx); err != nil {
		return fmt.Errorf("Wait: %v", err)
	}

	fmt.Fprintf(w, "Model deployed.\n")

	return nil
}

Java

在试用此示例之前，请按照客户端库页面中与此编程语言对应的设置说明执行操作。

在 Cloud Shell 中打开 在 GitHub 上查看 反馈

import com.google.api.gax.longrunning.OperationFuture;
import com.google.cloud.automl.v1.AutoMlClient;
import com.google.cloud.automl.v1.DeployModelRequest;
import com.google.cloud.automl.v1.ModelName;
import com.google.cloud.automl.v1.OperationMetadata;
import com.google.protobuf.Empty;
import java.io.IOException;
import java.util.concurrent.ExecutionException;

class DeployModel {

  public static void main(String[] args)
      throws IOException, ExecutionException, InterruptedException {
    // TODO(developer): Replace these variables before running the sample.
    String projectId = "YOUR_PROJECT_ID";
    String modelId = "YOUR_MODEL_ID";
    deployModel(projectId, modelId);
  }

  // Deploy a model for prediction
  static void deployModel(String projectId, String modelId)
      throws IOException, ExecutionException, InterruptedException {
    // Initialize client that will be used to send requests. This client only needs to be created
    // once, and can be reused for multiple requests. After completing all of your requests, call
    // the "close" method on the client to safely clean up any remaining background resources.
    try (AutoMlClient client = AutoMlClient.create()) {
      // Get the full path of the model.
      ModelName modelFullId = ModelName.of(projectId, "us-central1", modelId);
      DeployModelRequest request =
          DeployModelRequest.newBuilder().setName(modelFullId.toString()).build();
      OperationFuture<Empty, OperationMetadata> future = client.deployModelAsync(request);

      future.get();
      System.out.println("Model deployment finished");
    }
  }
}

Node.js

在试用此示例之前，请按照客户端库页面中与此编程语言对应的设置说明执行操作。

在 Cloud Shell 中打开 在 GitHub 上查看 反馈

/**
 * TODO(developer): Uncomment these variables before running the sample.
 */
// const projectId = 'YOUR_PROJECT_ID';
// const location = 'us-central1';
// const modelId = 'YOUR_MODEL_ID';

// Imports the Google Cloud AutoML library
const {AutoMlClient} = require('@google-cloud/automl').v1;

// Instantiates a client
const client = new AutoMlClient();

async function deployModel() {
  // Construct request
  const request = {
    name: client.modelPath(projectId, location, modelId),
  };

  const [operation] = await client.deployModel(request);

  // Wait for operation to complete.
  const [response] = await operation.promise();
  console.log(`Model deployment finished. ${response}`);
}

deployModel();

PHP

在试用此示例之前，请按照客户端库页面中与此编程语言对应的设置说明执行操作。

在 GitHub 上查看 反馈

use Google\Cloud\AutoMl\V1\AutoMlClient;

/** Uncomment and populate these variables in your code */
// $projectId = '[Google Cloud Project ID]';
// $location = 'us-central1';
// $modelId = 'my_model_id_123';

$client = new AutoMlClient();

try {
    // get full path of model
    $formattedName = $client->modelName(
        $projectId,
        $location,
        $modelId
    );

    $operationResponse = $client->deployModel($formattedName);
    $operationResponse->pollUntilComplete();
    if ($operationResponse->operationSucceeded()) {
        $result = $operationResponse->getResult();
        printf('Model deployed.' . PHP_EOL);
    } else {
        $error = $operationResponse->getError();
        // handleError($error)
    }
} finally {
    $client->close();
}

Python

在试用此示例之前，请按照客户端库页面中与此编程语言对应的设置说明执行操作。

在 GitHub 上查看 反馈

from google.cloud import automl

# TODO(developer): Uncomment and set the following variables
# project_id = "YOUR_PROJECT_ID"
# model_id = "YOUR_MODEL_ID"

client = automl.AutoMlClient()
# Get the full path of the model.
model_full_id = client.model_path(project_id, "us-central1", model_id)
response = client.deploy_model(name=model_full_id)

print(f"Model deployment finished. {response.result()}")

Ruby

在试用此示例之前，请按照客户端库页面中与此编程语言对应的设置说明执行操作。

在 GitHub 上查看 反馈

require "google/cloud/automl"

project_id = "YOUR_PROJECT_ID"
model_id = "YOUR_MODEL_ID"

client = Google::Cloud::AutoML.auto_ml

# Get the full path of the dataset
model_full_id = client.model_path project: project_id,
                                  location: "us-central1",
                                  model: model_id

operation = client.deploy_model name: model_full_id

# Wait until the long running operation is done
operation.wait_until_done!

puts "Model deployment finished."

更新模型的节点编号

如果您拥有经过训练的已部署模型，可以更新部署模型的节点数量，以响应您的特定流量。例如，如果您遇到的每秒查询次数 (QPS) 高于预期。

您可以在不必首先取消部署的情况下更改此节点数量。更新部署将更改节点数量，而不会中断您处理的预测流量。

节点数将受“配额”页面上列出的配额限制；默认情况下，一个用户最多可以获得 10 个节点。

网页界面

1. 在 Vision Dashboard中，选择左侧导航栏中的模型标签页（带有灯泡图标）以显示可用的模型。

   如需查看其他项目的模型，请从标题栏右上角的下拉列表中选择该项目。

2. 选择已部署的经过训练的模型。
3. 选择标题栏正下方的测试和使用标签页。

4. 页面顶部的框中将显示一条消息，其内容是“您的模型已部署，可用于在线预测请求”。 选择此文本旁边的更新部署选项。

   
5. 在打开的更新部署窗口中，从列表中选择用于部署模型的新节点编号。节点编号显示其估算的每秒预测查询次数 (QPS)。

6. 从列表中选择新节点编号后，选择更新部署以更新用于部署模型的节点编号。

   
7. 您将返回到测试和使用窗口，此时您会看到显示“正在部署模型…”的文本框。
   注意：这与模型的初始部署类似。
   

8. 模型成功部署到新节点编号后，您会在与项目关联的地址处收到一封电子邮件。

REST 和命令行

最初用于部署模型的方法与更改已部署模型的节点编号的方法相同。

在使用下面的任何请求数据之前，请先进行以下替换：

• project-id：您的 GCP 项目 ID。
• model-id：您的模型的 ID（从创建模型时返回的响应中获取）。此 ID 是模型名称的最后一个元素。 例如：
  • 模型名称：projects/project-id/locations/location-id/models/IOD4412217016962778756
  • 模型 ID：IOD4412217016962778756

字段注意事项：

• nodeCount - 需要在其上部署模型的节点的数量。值必须介于 1 到 100 之间，包括 1 和 100。节点是机器资源的抽象，可处理模型的 qps_per_node 中指定的每秒在线预测查询次数 (QPS)。

注意：节点数量将受配额页面上列出的配额的限制；默认情况下，用户最多可以获得 10 个节点。

HTTP 方法和网址：

POST https://automl.googleapis.com/v1/projects/project-id/locations/us-central1/models/model-id:deploy

请求 JSON 正文：

{
  "imageClassificationModelDeploymentMetadata": {
    "nodeCount": 2
  }
}

如需发送请求，请选择以下方式之一：

curl

注意：请确保已将 GOOGLE_APPLICATION_CREDENTIALS 环境变量设置为您的服务帐号私钥文件路径。

将请求正文保存在名为 request.json 的文件中，然后执行以下命令：

curl -X POST \
-H "Authorization: Bearer "$(gcloud auth application-default print-access-token) \
-H "Content-Type: application/json; charset=utf-8" \
-d @request.json \
https://automl.googleapis.com/v1/projects/project-id/locations/us-central1/models/model-id:deploy

PowerShell

注意：请确保已将 GOOGLE_APPLICATION_CREDENTIALS 环境变量设置为您的服务帐号私钥文件路径。

将请求正文保存在名为 request.json 的文件中，然后执行以下命令：

$cred = gcloud auth application-default print-access-token
$headers = @{ "Authorization" = "Bearer $cred" }

Invoke-WebRequest `
  -Method POST `
  -Headers $headers `
  -ContentType: "application/json; charset=utf-8" `
  -InFile request.json `
  -Uri "https://automl.googleapis.com/v1/projects/project-id/locations/us-central1/models/model-id:deploy" | Select-Object -Expand Content

您应该会看到类似如下所示的输出。可以使用操作 ID 来获取任务的状态。如需查看示例，请参阅处理长时间运行的操作。

{
  "name": "projects/project-id/locations/us-central1/operations/operation-id",
  "metadata": {
    "@type": "type.googleapis.com/google.cloud.automl.v1.OperationMetadata",
    "createTime": "2019-08-07T22:00:20.692109Z",
    "updateTime": "2019-08-07T22:00:20.692109Z",
    "deployModelDetails": {}
  }
}

您可以通过以下 HTTP 方法和网址获取操作的状态：

GET https://automl.googleapis.com/v1/projects/project-id/locations/us-central1/operations/operation-id

操作完成的状态将类似于以下内容：

{
  "name": "projects/project-id/locations/us-central1/operations/operation-id",
  "metadata": {
    "@type": "type.googleapis.com/google.cloud.automl.v1.OperationMetadata",
    "createTime": "2019-06-21T16:47:21.704674Z",
    "updateTime": "2019-06-21T17:01:00.802505Z",
    "deployModelDetails": {}
  },
  "done": true,
  "response": {
    "@type": "type.googleapis.com/google.protobuf.Empty"
  }
}

Go

在试用此示例之前，请按照客户端库页面中与此编程语言对应的设置说明执行操作。

在 GitHub 上查看 反馈

import (
	"context"
	"fmt"
	"io"

	automl "cloud.google.com/go/automl/apiv1"
	automlpb "google.golang.org/genproto/googleapis/cloud/automl/v1"
)

// visionClassificationDeployModelWithNodeCount deploys a model with node count.
func visionClassificationDeployModelWithNodeCount(w io.Writer, projectID string, location string, modelID string) error {
	// projectID := "my-project-id"
	// location := "us-central1"
	// modelID := "ICN123456789..."

	ctx := context.Background()
	client, err := automl.NewClient(ctx)
	if err != nil {
		return fmt.Errorf("NewClient: %v", err)
	}
	defer client.Close()

	req := &automlpb.DeployModelRequest{
		Name: fmt.Sprintf("projects/%s/locations/%s/models/%s", projectID, location, modelID),
		ModelDeploymentMetadata: &automlpb.DeployModelRequest_ImageClassificationModelDeploymentMetadata{
			ImageClassificationModelDeploymentMetadata: &automlpb.ImageClassificationModelDeploymentMetadata{
				NodeCount: 2,
			},
		},
	}

	op, err := client.DeployModel(ctx, req)
	if err != nil {
		return fmt.Errorf("DeployModel: %v", err)
	}
	fmt.Fprintf(w, "Processing operation name: %q\n", op.Name())

	if err := op.Wait(ctx); err != nil {
		return fmt.Errorf("Wait: %v", err)
	}

	fmt.Fprintf(w, "Model deployed.\n")

	return nil
}

Java

在试用此示例之前，请按照客户端库页面中与此编程语言对应的设置说明执行操作。

在 Cloud Shell 中打开 在 GitHub 上查看 反馈

import com.google.api.gax.longrunning.OperationFuture;
import com.google.cloud.automl.v1.AutoMlClient;
import com.google.cloud.automl.v1.DeployModelRequest;
import com.google.cloud.automl.v1.ImageClassificationModelDeploymentMetadata;
import com.google.cloud.automl.v1.ModelName;
import com.google.cloud.automl.v1.OperationMetadata;
import com.google.protobuf.Empty;
import java.io.IOException;
import java.util.concurrent.ExecutionException;

class VisionClassificationDeployModelNodeCount {

  static void visionClassificationDeployModelNodeCount()
      throws InterruptedException, ExecutionException, IOException {
    // TODO(developer): Replace these variables before running the sample.
    String projectId = "YOUR_PROJECT_ID";
    String modelId = "YOUR_MODEL_ID";
    visionClassificationDeployModelNodeCount(projectId, modelId);
  }

  // Deploy a model for prediction with a specified node count (can be used to redeploy a model)
  static void visionClassificationDeployModelNodeCount(String projectId, String modelId)
      throws IOException, ExecutionException, InterruptedException {
    // Initialize client that will be used to send requests. This client only needs to be created
    // once, and can be reused for multiple requests. After completing all of your requests, call
    // the "close" method on the client to safely clean up any remaining background resources.
    try (AutoMlClient client = AutoMlClient.create()) {
      // Get the full path of the model.
      ModelName modelFullId = ModelName.of(projectId, "us-central1", modelId);
      ImageClassificationModelDeploymentMetadata metadata =
          ImageClassificationModelDeploymentMetadata.newBuilder().setNodeCount(2).build();
      DeployModelRequest request =
          DeployModelRequest.newBuilder()
              .setName(modelFullId.toString())
              .setImageClassificationModelDeploymentMetadata(metadata)
              .build();
      OperationFuture<Empty, OperationMetadata> future = client.deployModelAsync(request);

      future.get();
      System.out.println("Model deployment finished");
    }
  }
}

Node.js

在试用此示例之前，请按照客户端库页面中与此编程语言对应的设置说明执行操作。

在 Cloud Shell 中打开 在 GitHub 上查看 反馈

/**
 * TODO(developer): Uncomment these variables before running the sample.
 */
// const projectId = 'YOUR_PROJECT_ID';
// const location = 'us-central1';
// const modelId = 'YOUR_MODEL_ID';

// Imports the Google Cloud AutoML library
const {AutoMlClient} = require('@google-cloud/automl').v1;

// Instantiates a client
const client = new AutoMlClient();

async function deployModelWithNodeCount() {
  // Construct request
  const request = {
    name: client.modelPath(projectId, location, modelId),
    imageClassificationModelDeploymentMetadata: {
      nodeCount: 2,
    },
  };

  const [operation] = await client.deployModel(request);

  // Wait for operation to complete.
  const [response] = await operation.promise();
  console.log(`Model deployment finished. ${response}`);
}

deployModelWithNodeCount();

PHP

在试用此示例之前，请按照客户端库页面中与此编程语言对应的设置说明执行操作。

在 GitHub 上查看 反馈

use Google\Cloud\AutoMl\V1\AutoMlClient;
use Google\Cloud\AutoMl\V1\ImageClassificationModelDeploymentMetadata;

/** Uncomment and populate these variables in your code */
// $projectId = '[Google Cloud Project ID]';
// $location = 'us-central1';
// $modelId = 'my_model_id_123';

$client = new AutoMlClient();

try {
    // get full path of model
    $formattedName = $client->modelName(
        $projectId,
        $location,
        $modelId
    );

    // set prediction node count
    $metadata = (new ImageClassificationModelDeploymentMetadata())
        ->setNodeCount(2);
    $args = ['imageClassificationModelDeploymentMetadata' => $metadata];

    $operationResponse = $client->deployModel($formattedName, $args);
    $operationResponse->pollUntilComplete();
    if ($operationResponse->operationSucceeded()) {
        $result = $operationResponse->getResult();
        printf('Model deployed.' . PHP_EOL);
    } else {
        $error = $operationResponse->getError();
        // handleError($error)
    }
} finally {
    $client->close();
}

Python

在试用此示例之前，请按照客户端库页面中与此编程语言对应的设置说明执行操作。

在 GitHub 上查看 反馈

from google.cloud import automl

# TODO(developer): Uncomment and set the following variables
# project_id = "YOUR_PROJECT_ID"
# model_id = "YOUR_MODEL_ID"

client = automl.AutoMlClient()
# Get the full path of the model.
model_full_id = client.model_path(project_id, "us-central1", model_id)

# node count determines the number of nodes to deploy the model on.
# https://cloud.google.com/automl/docs/reference/rpc/google.cloud.automl.v1#imageclassificationmodeldeploymentmetadata
metadata = automl.ImageClassificationModelDeploymentMetadata(node_count=2)

request = automl.DeployModelRequest(
    name=model_full_id, image_classification_model_deployment_metadata=metadata
)
response = client.deploy_model(request=request)

print("Model deployment finished. {}".format(response.result()))

Ruby

在试用此示例之前，请按照客户端库页面中与此编程语言对应的设置说明执行操作。

在 GitHub 上查看 反馈

require "google/cloud/automl"

project_id = "YOUR_PROJECT_ID"
model_id = "YOUR_MODEL_ID"

client = Google::Cloud::AutoML.auto_ml

# Get the full path of the model.
model_full_id = client.model_path project: project_id,
                                  location: "us-central1",
                                  model: model_id
metadata = { node_count: 2 }

operation = client.deploy_model name: model_full_id,
                                mage_classification_model_deployment_metadata: metadata

# Wait until the long running operation is done
operation.wait_until_done!

puts "Model deployment finished."