If I have a function,
import torch
layer = torch.nn.Conv2d(1, 6, 5)
When torch.nn.Conv2d runs, how to get the C++ file of this system call?
If I have a function,
import torch
layer = torch.nn.Conv2d(1, 6, 5)
When torch.nn.Conv2d runs, how to get the C++ file of this system call?
You could run the python
executable via gdb --args python script.py args
and use breakpoints to check the backend functions or directly step through the execution.