无法在Spark中为决策树保存PipelineModel

from pyspark.ml.classification import DecisionTreeClassifier
from pyspark.ml.evaluation import MulticlassClassificationEvaluator

# 创建决策树分类器
dt = DecisionTreeClassifier(labelCol="label", featuresCol="features")

# 训练模型
model = dt.fit(train_data)

# 保存模型
model.save("path/to/save/model")

# 加载模型
loaded_model = DecisionTreeModel.load("path/to/save/model")

# 使用加载的模型进行预测
predictions = loaded_model.transform(test_data)

# 评估模型
evaluator = MulticlassClassificationEvaluator(labelCol="label", predictionCol="prediction", metricName="accuracy")
accuracy = evaluator.evaluate(predictions)