List All Trainable Variables in PyTorch – PyTorch Tutorial

By | June 13, 2024

In this tutorial, we will introduce how to list all trainable variables in pytorch, which is very useful when you are fine-tuning a model.

How to list all trainable variables in pytorch?

It is very easy to implement, here is an example:

vars = system.parameters()
for v in vars:
    if v.requires_grad:
        print(v.numel(), v.requires_grad)

Here system is a pytorch model, if a pytorch variable is trainable, its requires_grad will be True.

Run this code, you may view:

List All Trainable Variables in PyTorch - PyTorch Tutorial

Moreover, if we want to show the parameter name, we can do as this:

vars = system.named_parameters()
for n, k in vars:
    if k.requires_grad:
        print(n, k.requires_grad)

Then, we may see:

pytorch show all trainable variable names

Leave a Reply

Your email address will not be published. Required fields are marked *