MasteringPyTorchV2 icon indicating copy to clipboard operation
MasteringPyTorchV2 copied to clipboard

Incorrect GoogLeNet forward pass and layer naming

Open gangula-karthik opened this issue 9 months ago • 0 comments

The provided implementation of GoogLeNet contains several mistakes:

  • Wrong usage of undefined layers like a4, b4, etc. instead of the defined im3, im4, etc.
  • Incorrect data flow: output of im1 is not passed to im2; instead, the original input is reused.
  • Typographical errors: using maxpool instead of max_pool, and avgerage_pool instead of average_pool.

gangula-karthik avatar Apr 27 '25 09:04 gangula-karthik