Encoding Decoder Models

We looked at like waiting stuff by duplicates to simplify the process. A lot of the time there's actually quite big divergence between the image vector and the and the text vector. Data points where the image and caption could end up being very similar in their embedding are going to tell the model about specific things. Things that differ strongly from their caption and image help the model to generalize.

Transcript

Play full episode

Remember Everything You Learn from Podcasts

Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.