训练VGG-16需要多少个时期 [英] How many epochs does it take to train VGG-16
问题描述
我正在使用包含3k图像的数据集从头开始训练VGG-16模型.我在没有任何GPU的情况下使用Tensorflow平台和8 cpus.
I'm training a VGG-16 model from scratch using a dataset containing 3k images. I use Tensorflow platform and 8 cpus without any gpu.
培训率-0.01,
重量衰减-0.0005,
Weight decay - 0.0005,
动量-0.9,
批量大小-64,
我已经训练了大约三天.但是训练的准确性并没有改变,在20个时期后大约为15%-20%.有人可以给我一些提高准确性的提示吗?
I've kept training for about three days. But the training accuracy has been unchanged, around 15%-20% after 20 epochs. Could anyone give me some hints to improve the accuracy?
推荐答案
似乎我使用了太大的学习率.或体重减轻不能像预期的那样起作用.将超级参数更改为
It seems like I have used too large learning rate. Or weight decay does not work as it promises. After I changed my hyper parameters into,
培训率-0.001,
重量衰减-0,
动量-0.9,
批量大小-64.
一切都很好.现在,经过25个纪元后,我可以获得90%左右的准确度.仅供参考.
Eveything is good. Now I can get accuracy around 90 percent after 25 epochs. Just for your information.
这篇关于训练VGG-16需要多少个时期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!