- PyTorch broadcasting rules — read this carefully before doing tensor arithmetic across dimensions of different shapes (e.g. row/column normalization). Getting
keepdimwrong here causes silent, hard-to-spot bugs.
| Name | Name | Last commit date | ||
|---|---|---|---|---|
keepdim wrong here causes silent, hard-to-spot bugs.