PyTorch torch.set_deterministic_debug_mode Function
Pytorch torch Reference Manual
torch.set_deterministic_debug_modeThis function is used to set the deterministic debug mode in PyTorch. When set to a non-zero value, PyTorch will throw an error or warning when a non-deterministic algorithm is used.
Function Definition
torch.set_deterministic_debug_mode(debug_mode)
Parameter Description
debug_mode: Debug mode, optional values:0: Off (default)1: Enable strict mode2: Enable warning mode
Usage Example
Example
import torch
# Set the deterministic debug mode to strict mode
# This will throw an error when non-deterministic operations are used
torch.set_deterministic_debug_mode(1)
# Set to warning mode
torch.set_deterministic_debug_mode(2)
# Turn off debug mode
torch.set_deterministic_debug_mode(0)
print("Deterministic debug mode has been set")
# Set the deterministic debug mode to strict mode
# This will throw an error when non-deterministic operations are used
torch.set_deterministic_debug_mode(1)
# Set to warning mode
torch.set_deterministic_debug_mode(2)
# Turn off debug mode
torch.set_deterministic_debug_mode(0)
print("Deterministic debug mode has been set")
The output is:
Deterministic debug mode has been set
Other Extensions