管道 - 将数据生成到训练模型 [英] pipeline - data generation to trained model

问题描述

大家好，

ML管道对于集成用python编写的阶段非常有用。在我的情况下，我在开始时有一个额外的服务，它在程序上生成训练数据。

The ML pipeline looks useful for integrating stages that are written in python. In my case I have an additional service at the beginning which is generating procedurally the data for training.

我应该在管道的所有阶段使用ML管道（数据生成 - 三角形 - 发布模型），包括生成数据和调用外部服务的开头的步骤？或者有一个不同的服务建议应该将
全部放在一起，比如数据工厂或服务结构，我应该坚持使用ML管道只用于基于python的训练阶段？

Should I use the ML pipeline for all the stages of the pipeline (data generation - trining - publish model), including the steps at the beginning that generate data and invoke external services? Or there is a different service suggested that should bring all together like Data factory, or Service fabric and I should stick with ML pipeline only for the python based training stages?

谢谢，

伊曼纽尔

 

Emanuel Shalev

Emanuel Shalev

推荐答案

您好，

是，如果您希望从数据生成流程到模型创建Azure ML管道是最适合的，因为它是为这个用例而设计的。您甚至可以查看

azure pipelines 如果您一直在考虑更新代码，因为它将训练和创建新的模型，如CI / CD管道。 

Yes, If you are looking to have a flow from data generation to model creation Azure ML pipeline is a best fit as it is designed for this use case. You can even look at azure pipelines if you are constantly looking at updating your code as it will train and create new models like a CI/CD pipeline. 

这篇关于管道 - 将数据生成到训练模型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IT屋！