Did you check your GPU memory during the run or just after it crashed? Because that spike could also be from the log writing itself if wandb was buffering a lot of data on the same card. Next time try watching nvidia-smi in a second terminal while it trains, you'll usually see the memory creep up right before things go sideways.

Try resting the dough for 20 minutes instead of an hour and see if that changes your mind. Cheap flour needs all the help it can get, but your Dutch oven is doing way more work than autolyse ever will.

Best move is to nod and keep working till they get to the point.