> 关于线上环境不提供GPU的一些疑问

这是否意味着多数深度学习的方法不能使用，而且每个数据集给的时间预算和计算资源其实并不多，这意味着要将方案限制在传统的机器学习和时序预测方法上吧？

Dear Participant,

Many thanks for your participation!

We have no limitations on methods. You can try any methods you like e.g. deep neural networks to solve this problem using limited resources and time budget.

The time budget and computation resources are the same for all the methods. In real-world scenarios, we always need consider the cost efficiency of our proposed methods.

Moreover, we've already extended the time budget and computation resources a lot comparing to the cost of baseline method. We might extend the resources more in the next AutoSeries competitions if needed, but not in this competition.

Good luck! Thanks again for your participation!

Best,
AutoSeries Competition Organizers