What was the validation accuracy before any training started?
What was the validation accuracy before any training started?: a task in SmolDataEnvs-harbor-train-sandbox: 📊 SmolDataEnvs: Harbor (train) (Harbor dataset). Files (in /home/user/input, no subfolders): - cifar-10-python.tar.gz
The task
Files (in /home/user/input, no subfolders): - cifar-10-python.tar.gz