A Python utility to reload a loop body from source on each iteration without loosing state
Useful for editing PyTorch code during training. This lets you e.g. add logging, print statistics or save the model without restarting the training and, therefore, without loosing the training progress.
Please make sure you're using Python 3, then install via pip:
pip install reloading
Simply wrap the iterator in a for loop with reloading, e.g.
import reloading from reloading
for i in reloading(range(10)):
# here could be your training loop
print(i)This will reload the loop body from the source file before each iteration.