PyTorch torch.cuda.seed_all Function
Pytorch torch Reference Manual
torch.cuda.seed_allIt is a function in PyTorch used to set the random seed for all CUDA devices. Sets the random seed for all CUDA devices (functionally identical to manual_seed_all).
Function Definition
torch.cuda.seed_all()
Usage Example
Example
import torch
# Set the random seed for all CUDA devices
if torch.cuda.is_available():
torch.cuda.seed_all()
x = torch.randn(3, 3).cuda()
print(f"Random tensor: {x}")
print(f"CUDA device count: {torch.cuda.device_count()}")
else:
print("CUDA not available")
# Set the random seed for all CUDA devices
if torch.cuda.is_available():
torch.cuda.seed_all()
x = torch.randn(3, 3).cuda()
print(f"Random tensor: {x}")
print(f"CUDA device count: {torch.cuda.device_count()}")
else:
print("CUDA not available")
Other Extensions