search

davidsonic / Interpretable_CNNs_via_Feedforward_Design

Official Implementation of Interpretable Convolutional Neural Networks via Feedforward Design Arxiv: https://arxiv.org/abs/1810.02786
11 stars 8 forks source link

Why the test acc is so low? #1

Open lianqiyan opened 5 years ago

lianqiyan commented 5 years ago

I run the following scripts python Getkernel.py python Getfeature.py python Getweight.py python mnist_test.py But the testing acc is only 0.0974.

minzhang-1 commented 5 years ago

There is a small problem in saab.py, have solved.

leader120 commented 4 years ago

I run the following scripts python Getkernel.py python Getfeature.py python Getweight.py python mnist_test.py But the testing acc is only 0.0974.

可以告诉我一下您当时运行这个代码有没有出现内存占用过大的错误

minzhang-1 commented 4 years ago

If you run this code on your own laptop, it might have memory error. It depends on the size of your memory. You can use the parameters we provide or you can modify the code to batch based to fit in your memory. For example, you can change the pca to incremental pca so that you can fit the data by batch.

leader120 commented 4 years ago

If you run this code on your own laptop, it might have memory error. It depends on the size of your memory. You can use the parameters we provide or you can modify the code to batch based to fit in your memory. For example, you can change the pca to incremental pca so that you can fit the data by batch.

Thanks,but the feat.pkl not in your code,i should run the Getfeature.py to get the feat.pkl,then memory error have happened , you can give me the feat.pkl? and i have a quenstion ,When you do experiments, how much memory does the computer have？