What values do the the three : : : take on in this code?

{Moderator edit: code removed}

In the lines of code weights = W[:, :, :, c]
biases = b[:, :, :, c] what takes the place of :,:,: is it i h and w ?

Hi @Stephano_Cotsoradis ,
The three colons (:::), also known as the ellipsis operator, means that the tensor W will be indexed along all of its dimensions, except for the last dimension, which is indexed by the variable C.

In the assignment the colons are separated by commas… So I don’t think that’s an ellipsis.
The colon is a wildcard index - meaning to fetch everything.

Also I’m deleting the code from the post, because posting your code isn’t allowed by the Code of Conduct.

Oh okay sorry I did not know that.

No problem.

In that situation would m take the place of the first : and then h the second : and w the third : and then c is just the layer ?

Not really. Those are the definitions of what each of those dimensions are used for, but you would not use the variable names because they would only give you one specific value.

I realize I am not quite using the correct terminology, but it would take the value of h for that iteration and then the value of h would be different when the filter moves down correct ?

Sorry, I don’t understand your question.
The colon when used as an index will match all values for that index. You don’t get any selection of values.

that answers it for me thanks !