Search⌘ K
Home
Courses
Deep Learning with PyTorch Step-by-Step: Part I - Fundamentals

Loading Data, Devices, and CUDA

Explore converting Numpy arrays into PyTorch tensors while managing data on CPU and GPU devices. Understand how to check GPU availability with CUDA, transfer tensors between devices, and avoid common conversion errors to prepare your data effectively for training deep learning models.

We'll cover the following...

Conversions between Numpy and PyTorch

It is time to start converting our Numpy code to PyTorch. We will start with the training data; that is, our x_train and y_train arrays.

The “as_tensor” method

”How do we go from Numpy’s arrays to PyTorch’s tensors?”

The as_tensor method preserves the type of the array, which can also be seen in the code below:

Python 3.5
# Using as_tensor function to convert Numpy array to PyTorch tensor
x_train_tensor = torch.as_tensor(x_train)
print(x_train.dtype, x_train_tensor.dtype)

You can also easily cast it to a different type like a lower precision (32-bit) float, which will occupy less space in memory using the float() method:

Python 3.5
# Using float function to change data types 
x_train_tensor = torch.as_tensor(x_train)
float_tensor = x_train_tensor.float() # (going from float64 to float32) 
print(float_tensor.dtype)
...

IMPORTANT: Both as_tensor() and from_numpy() return a tensor that shares the underlying data with the original Numpy array. Similar to what happened when we used view() in the previous lesson, if you modify the original Numpy array, you are modifying the corresponding PyTorch tensor too and vice-versa. ...