如何将Scikit Learn OneVsRestClassifier预测方法输出转换为google cloud ML的密集数组？

import numpy as np
from sklearn.multiclass import OneVsRestClassifier
from sklearn.svm import SVC
from google.cloud import storage
from google.cloud import ml

# 加载训练数据
X_train = np.load('train_data.npy')
y_train = np.load('train_labels.npy')

# 加载模型
model = OneVsRestClassifier(SVC())
model.fit(X_train, y_train)

# 加载测试数据
X_test = np.load('test_data.npy')

# 进行预测
y_pred = model.predict(X_test)

# 转换为密集数组
dense_array = y_pred.toarray()

# 设置Google Cloud Storage的bucket和文件名
bucket_name = 'your_bucket_name'
file_name = 'dense_array.npy'

# 上传密集数组到Google Cloud Storage
client = storage.Client()
bucket = client.get_bucket(bucket_name)
blob = bucket.blob(file_name)
blob.upload_from_filename('dense_array.npy')

# 设置Google Cloud ML的模型和版本
project_id = 'your_project_id'
model_name = 'your_model_name'
version_name = 'your_version_name'

# 创建Cloud ML客户端
ml_client = ml.Client(project_id)

# 获取模型版本
model_version = ml_client.get_model(model_name).versions.get(version_name)

# 设置预测请求
input_data = {'dense_array': 'gs://{}/{}'.format(bucket_name, file_name)}
output_path = 'gs://{}/{}'.format(bucket_name, 'predictions')
prediction_input = ml.Input(input_data, 'jsonl')
prediction_output = ml.Output(output_path, 'jsonl')

# 发送预测请求
job = model_version.predict(prediction_input, output=prediction_output)
job.wait()