Skip to content

Image Captioning

Rather than instructing the RNN to sample text at random, we are conditioning that sampling by the output of the CNN

Forward Pass

Backward pass

  • If you start with pre-trained CNN, only backprop for the RNN
  • Else, backprop through the RNN and CNN
Last Updated: 2024-12-26 ; Contributors: AhmedThahir, web-flow

Comments