Possible bug in WideResnet implementation

TimDettmers / sparse_learning

Sparse learning library and sparse momentum resources.

MIT License

377 stars 45 forks source link

Possible bug in WideResnet implementation #13

Closed micronet-challenge-submissions closed 4 years ago

micronet-challenge-submissions commented 4 years ago

https://github.com/TimDettmers/sparse_learning/blob/6b978515da7b9e420f518711825d23cc394f8864/sparselearning/models.py#L353

Hi Tim,

Looks like the depth is not used by the model. Is this intended?

TimDettmers commented 4 years ago

It is a bit difficult to spot, but the depth variable is used for the n variable which is then used in the NetworkBlock.

Please let me know if you find other possible bugs!

evcu commented 4 years ago

Hi Tim,

Sorry that I opened up the bug using the wrong account. Thanks for your response. It make sense, I missed that the NetworkBlock uses n to decide n_layers. Makes sense!