Skip to main content

Deep Learning Frameworks

All commands in Windows Command Prompt, and this will work for manual Python installation only.​

Check for GPU Physical presence:​

nvidia-smi

TensorFlow​

import tensorflow as tf; print(tf.config.list_physical_devices())

This will return a list like [PhysicalDevice(name='/physical_device:CPU:0', device_type='CPU')], check if GPU device is found, if yes then use the device number (ex: CPU:0) in code.

import tensorflow as tf
tf.config.list_physical_devices('GPU')

# sample output -- [PhysicalDevice(name='/physical_device:GPU:0', device_type='GPU'),
# PhysicalDevice(name='/physical_device:GPU:1', device_type='GPU')]

PyTorch​

import torch; print(torch.cuda.is_available())

This will return True if GPU is available, else False