在 Java 中的 Gemini Pro 版本上使用 Bearer Api 令牌

Question

背景

在尝试在 java 项目中使用 API 令牌（而不是登录）向 Gemini Pro 模型提问时，我在使用不记名令牌（原始 API 令牌）访问模型时遇到了一些困难。基于this答案，我尝试过：

public static void callViaBearerApiToken(
      String projectId, String location, String modelName, String bearerToken)
      throws ClientProtocolException, IOException {
    // TO get the endPointId type: gcloud ai endpoints list --region=${LOCATION} --uri in cli.
    String endpointId = "https://us-central1-aiplatform.googleapis.com/";
    // Define the endpoint URL
    String endpointUrl =
        "https://"
            + location
            + "-aiplatform.googleapis.com/v1/projects/"
            + projectId
            + "/locations/"
            + location
            // + "/endpoints/"
            // + endpointId
            + ":predict";
    System.out.println("endpointUrl=\n" + endpointUrl);
    String someQuestion = "Can you say hello world?";

    // Define the request body
    JsonObject requestBody = new JsonObject();
    JsonObject instance = new JsonObject();
    instance.addProperty("content", someQuestion);
    requestBody.add("instances", instance);

    // Convert the request body to JSON
    String requestBodyJson = new Gson().toJson(requestBody);

    // Create an HTTP client and request
    HttpClient client = HttpClients.createDefault();
    HttpPost request = new HttpPost(endpointUrl);
    StringEntity requestEntity = new StringEntity(requestBodyJson);
    requestEntity.setContentType("application/json");
    request.setEntity(requestEntity);

    // Set the authentication header
    request.setHeader(HttpHeaders.AUTHORIZATION, bearerToken);

    // Send the request and get the response
    HttpResponse response = client.execute(request);
    HttpEntity responseEntity = response.getEntity();
    String responseBodyJson = EntityUtils.toString(responseEntity);

    // Get response headers
    org.apache.http.Header[] headers = response.getAllHeaders();
    for (org.apache.http.Header header : headers) {
        System.out.println(header.getName() + ": " + header.getValue());
    }
  
    // Get the status code
    int statusCode = response.getStatusLine().getStatusCode();
    System.out.println("Status Code: " + statusCode);


    // Deserialize the response body from JSON
    JsonObject responseBody = new Gson().fromJson(responseBodyJson, JsonObject.class);

    // Process the response
    System.out.println("response=\n" + responseBody);
    System.out.println("response again=\n" + responseBody.getAsString());
  }

然而，这会产生：

endpointUrl=
https://us-central1-aiplatform.googleapis.com/v1/projects/someProjectName/locations/us-central1:predict

Content-Type: text/html
Date: Sun, 11 Feb 2024 18:04:50 GMT
Server: scaffolding on HTTPServer2
Content-Length: 0
X-XSS-Protection: 0
X-Frame-Options: SAMEORIGIN
X-Content-Type-Options: nosniff
Alt-Svc: h3=":443"; ma=2592000,h3-29=":443"; ma=2592000

Status Code: 404

response=
null

Exception in thread "main" java.lang.NullPointerException: Cannot invoke "com.google.gson.JsonObject.getAsString()" because "responseBody" is null
    at com.doctestbot.prepare_calls.CallGoogle.callViaBearerApiToken(CallGoogle.java:81)
    at com.doctestbot.Main.main(Main.java:53)

使用以下命令生成 Bearer API 令牌后：

gcloud auth application-default print-access-token

问题

我怎样才能问“你能打个招呼吗？”对于 Java 中的 Gemini Pro 模型，使用 Bearer API 令牌作为唯一身份验证方式？

疑问

我认为我应该从谷歌云设置中的某个地方获取端点ID，因为大多数其他问题提到这会屏蔽它以保护隐私，然后输入：

gcloud ai endpoints list --region=us-central1 --uri

输出：

Using endpoint [https://us-central1-aiplatform.googleapis.com/]
Listed 0 items.

但是，在下面的示例中似乎没有必要手动指定端点。

注意

为了完整起见，我确保我可以使用具有正常身份验证的 Gemini Pro 模型，使用 this 示例：

import com.google.cloud.vertexai.VertexAI;
import com.google.cloud.vertexai.api.GenerateContentResponse;
import com.google.cloud.vertexai.generativeai.preview.ChatSession;
import com.google.cloud.vertexai.generativeai.preview.GenerativeModel;
import com.google.cloud.vertexai.generativeai.preview.ResponseHandler;
import java.io.IOException;

public class ChatDiscussion {

  public static void main(String[] args) throws IOException {
    // TODO(developer): Replace these variables before running the sample.
    String projectId = "your-google-cloud-project-id";
    String location = "us-central1";
    String modelName = "gemini-pro";

    chatDiscussion(projectId, location, modelName);
  }

  // Ask interrelated questions in a row using a ChatSession object.
  public static void chatDiscussion(String projectId, String location, String modelName)
      throws IOException {
    // Initialize client that will be used to send requests. This client only needs
    // to be created once, and can be reused for multiple requests.
    try (VertexAI vertexAI = new VertexAI(projectId, location)) {
      GenerateContentResponse response;

      GenerativeModel model = new GenerativeModel(modelName, vertexAI);
      // Create a chat session to be used for interactive conversation.
      ChatSession chatSession = new ChatSession(model);

      response = chatSession.sendMessage("Hello.");
      System.out.println(ResponseHandler.getText(response));

      response = chatSession.sendMessage("What are all the colors in a rainbow?");
      System.out.println(ResponseHandler.getText(response));

      response = chatSession.sendMessage("Why does it appear when it rains?");
      System.out.println(ResponseHandler.getText(response));
      System.out.println("Chat Ended.");
    }
  }
}

Answer 1

如果您想手动构建并发送 HTTP 请求（而不是让 Vertex AI SDK 为您完成），则选择 REST 选项卡将为您提供所需的所有信息。特别是，您将看到端点是：

https://<LOCATION>-aiplatform.googleapis.com/v1/projects/<PROJECT_ID>/locations/<LOCATION>/publishers/google/models/<MODEL_ID>:streamGenerateContent

它与您尝试过的不一样（这就是您得到 404 的原因）。

另请注意请求正文：

{
  "contents": {
    "role": <ROLE>,
    "parts": { "text": <TEXT> }
  },
  "safety_settings": {
    "category": <SAFETY_CATEGORY>,
    "threshold": <THRESHOLD>
  },
  "generation_config": {
    "temperature": <TEMPERATURE>,
    "topP": <TOP_P>,
    "topK": <TOP_K>,
    "candidateCount": 1,
    "maxOutputTokens": <MAX_OUTPUT_TOKENS>,
    "stopSequences": <STOP_SEQUENCES>
  }
}

它也与你尝试过的不同。