Last month I was processing 80,000 images for my project.
Colab dropped at 71,000.
I stared at the screen. Then restarted from zero. Third time that week.
I was so frustrated I stopped working for that day.
Then I built loopz.
python@loopz.track("my_job", save_every=100)
def process(image_path):
do_work(image_path)
process(all_images)
crash? just run again → resumes from exactly where it stopped ✅
That's the entire code change. Nothing else.
It saves your model weights, optimizer, random state — everything. Automatically.
*pip install loopz
*
GitHub → https://github.com/Shiv0087/loopz
If you ever lost hours to a crash, this is for you.
Top comments (0)