Typos in Convolutional Networks page

[anguyen8]

Hi Andrej,

Thanks for putting such a great resource about convnets and Deep Learning online. I believe this would be greatly helpful for many people including me.

I was reading the convnets page and found this minor issue that you might want to fix. The page says:

The Krizhevsky et al. architecture ...  Real-world example: Since (227 - 11)/4 + 1 = 55, we can see that ...

However, they actually use crops of 224x224. So the formula would be (224 - 11)/4 + 1 = 54.25, which is not an integer but I guess still valid configurations, right?

For the case of Zeiler et al. (Visualizing and Understanding convnets), this formula also gives a non-integer result: (224 - 7)/2 + 1 = 109.5..

Best,

Anh

[karpathy]

Hey Anh, thanks I should maybe address this explicitly in notes.

The Caffe model of AlexNet uses 227x227 images, as it should be. These dimensions make sense.

It's not clear to me how Alex made 224x224 images fit because based on the numbers he gives in the paper the dimensions don't work out. He padded the images somehow in unmentioned way.

[anguyen8]

@karpathy : Thanks for your response. Yes, I agree it seems like they must use paddings somewhere.