Calculating RAM size of network

1

I’d really like a call to print the RAM footprint of my nn.Module (including gradients) - does this exist?

I know that there will be more requirements for training data etc, but it does seem possible to calculate just the Module requirements deterministically.

Thanks.

2

This post (Finding model size) might help you.
If you want to print the RAM occupancy for each module, here’s an example snippet

for m in model.modules():
    for param in m.parameters():
        print(param.nelement() * param.element_size())
1 Like
3

Thanks - would that give KB?

4

It would be byte.
Divide the output with 1024 gives you KB

1 Like