问题描述
我使用Azure AutoML训练了多类分类模型。模型返回标签而不是具有与每个标签关联的概率的数组。如果我在Azure ML Studio中训练模型并部署服务,则模型的行为与预期的
相同,并返回每个类的概率。有没有办法指示使用AutoML训练的模型所需的输出类型?
I have trained a multiclass classification model using Azure AutoML. The model returns a label instead of an array with the probabilities associated with each label. If I train the model in Azure ML Studio and deploy the service, the model behaves as expected and returns the probabilities for each class. Is there a way to indicate the type of output required from a model trained using AutoML?
推荐答案
注意到AutoMLConfig类中的num_classes参数。你试过这个吗?
Noticed a num_classes parameter in the AutoMLConfig class. Did u try this?
- num_classes
分类实验的标签数据中的类数。
- num_classes
- int
Number of classes in the label data for a classification experiment.
这篇关于更改Azure AutoML分类模型结果的形状的文章就介绍到这了,希望我们推荐的答案对大家有所帮助,也希望大家多多支持!