How can i convert model's parameter data type to Double?

1

Default data type of parameters is Float.
I want to convert it to Double.
How can i do this?

2 Likes
2

You can convert your model to double by doing model.double().
Note that after this, you will need your input to be DoubleTensor.

9 Likes
3

Thanks very much! It’s very simple.

4

Convert Your Data to float instead … as it’s very dangerous to convert model type to Double … as GPUs are not good at double precision computation
Try:

X_train = torch.FloatTensor(X_train)

instead

1 Like