如何将云存储中的Tensorboard日志上传到VertexAI-Tensorboard？

以下是一个Python代码示例，展示如何从Google Cloud Storage下载Tensorboard日志，并将它们上传到Vertex AI-Tensorboard。

from google.cloud import storage
from google.cloud.aiplatform_v1beta1 import TensorboardServiceClient
from google.cloud.aiplatform_v1beta1.types import CreateTensorboardRequest, Tensorboard

# 从Google Cloud Storage下载Tensorboard日志并上传到Vertex AI-Tensorboard
def upload_logs_to_vertex_ai(storage_bucket_name, log_dir, tensorboard_resource_name):
    # 设置Google Cloud Storage客户端并下载Tensorboard日志
    client = storage.Client()
    bucket = client.get_bucket(storage_bucket_name)
    blob_list = list(bucket.list_blobs(prefix=log_dir))
    for blob in blob_list:
        # 将Tensorboard日志内容读取为字符串
        log_content = blob.download_as_string()
        # 创建Tensorboard对象用于上传日志
        tensorboard = Tensorboard(name=tensorboard_resource_name)
        request = CreateTensorboardRequest(parent="projects/my-project/locations/us-central1", tensorboard=tensorboard)
        with TensorboardServiceClient() as tensorboard_client:
            # 创建Tensorboard资源
            operation = tensorboard_client.create_tensorboard(request=request)
        print("Tensorboard resource created!")
        # 等待Tensorboard资源创建完成
        operation.result()
        # 上传Tensorboard日志
        with TensorboardServiceClient() as tensorboard_client:
            tensorboard_client.write_tensorboard_experiment_data(
                tensorboard_experiment=f"{tensorboard_resource_name}/experiments/my-experiment",
                data=log_content,
                content_type="text/plain",
            )
    print("Tensorboard logs uploaded to Vertex AI-Tensorboard!")

该示例演示了如何使用Google Cloud Storage客户端下载Tensorboard日志，并使用Vertex AI-Tensorboard API上传日志。首先，使用storage.Client()设置Google Cloud Storage客户端，并通过传递存储桶名称和日志目录，从云存储中下载Tensorboard日志。接下来，创建一个Tensorboard对象，并使用CreateTensorboardRequest对象