A bug that taught me more about PyTorch than years of using it(elanapearl.github.io)
426 points by bblcla | 78 comments
Blog post recounting how a training loss plateau turned out to be a PyTorch bug, traced through optimizer internals down to GPU kernels.
