wujiekd / Predicting-used-car-prices

阿里天池与Datawhale联合举办二手车价格预测比赛:优胜奖方案代码总结
99 stars 44 forks source link

神经网络数据与模型维度不同吧 #1

Open hellokevin96 opened 4 years ago

hellokevin96 commented 4 years ago

你好,看了你的代码深受启发。但也发现了一个问题:神经网络模型的输入是512, 但是真实数据维度不是512吧? 期待你的回复

wujiekd commented 4 years ago

嗯嗯,这个是keras设计网络结构其中一种非常直观的办法,不需要定义输入层,512是第一层FC的神经元个数,您也可以使用另外更有效的模型设计方法。希望可以帮助到您~

hellokevin96 commented 4 years ago

好的,谢谢

------------------ 原始邮件 ------------------ 发件人: "wujiekd"<notifications@github.com>; 发送时间: 2020年6月19日(星期五) 晚上8:00 收件人: "wujiekd/Predicting-used-car-prices"<Predicting-used-car-prices@noreply.github.com>; 抄送: "刘文斌"<1033704835@qq.com>; "Author"<author@noreply.github.com>; 主题: Re: [wujiekd/Predicting-used-car-prices] 神经网络数据与模型维度不同吧 (#1)

嗯嗯,这个是keras设计网络结构其中一种非常直观的办法,不需要定义输入层,512是第一层FC的神经元个数,您也可以使用另外更有效的模型设计方法。希望可以帮助到您~

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.