Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
jbschlosser
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
4 ms
·
1.
▲
by
jbschlosser
5y ago
FYI PyTorch now supports "same" padding for convolutions with stride=1; see the docs here: https://pytorch.org/docs/stable/generated/torch.nn.Conv2d.ht... The stride>1 case has been a bit more co
2.
▲
by
jbschlosser
5y ago
This is supported by PyTorch as well with nn.LazyLinear: https://pytorch.org/docs/stable/generated/torch.nn.LazyLinea... .