Rumor
A claimed PyTorch autograd zero-day could allow any computation during training
A user claims a regularization term can use model weights as RAM and SGD as a CPU clock.
TLDR
A user claims a supply-chain zero-day in PyTorch autograd lets a regularization term use a model’s weights as RAM and SGD as its CPU clock to run any computation during training. The user questions whether the issue is patchable.
Combined views
6.4K
4 Sources, first seen ago
