Dred

icon indicating a website link https://lingbai-kong.github.io/ icon indicating an email [email protected]

icon company ByteDance icon location Shanghai, China icon location TikTok, ByteDance | M.Sc. of ADMIS Lab, Tongji Univ.

Results 1 comments of Dred

[BUG Report]: duplicated variable names in layer components causes the loading weights failure.

Keeping two embedding layers separate can solve the problem. But the loss does not decrease without `StackLayers` (Not this example). It seems the embedding layers are not considered in the...