Parameter Initialization

English Version pytorch
1

https://d2l.ai/chapter_builders-guide/init-param.html

1 Like
2

For most of the examples, since the network was already initialized and computed on X in the first subsection, the example conditional if type(module) == nn.LazyLinear will not work since the data type was converted to nn.Linear, and if the network is not instantiated, then the net.apply(...) method will not work either

1 Like
3

Yes, this section seems to have some errors.