Quiz · 4 questions
🏋️ Training Data, State, and Checkpoints
Dataloaders, modes, saving, and evaluation
Level 0Curious
0 XP0/73 lessons0/11 achievements
0/120 XP to next level120 XP to go0% complete
Quiz
01Which two settings should ALWAYS be different between a training DataLoader and a validation DataLoader?
02What's the most common silent bug from forgetting
opt.zero_grad()?03What MUST be in a checkpoint if you want to resume training (not just inference)?
04Why do most teams accept that GPU training is not bitwise reproducible across runs, even with all seeds set?
Comments 0
🔔 Reply notifications (sign in)Sign in — Please sign in to comment.
No comments yet — be the first.