About this document
L15e TransformerDecoder by Afsheen hamid is a document available to read on EtoBox.
The document provides an overview of the decoder layer in transformer models, detailing key components such as input embeddings, positional encoding, and multi-headed attention mechanisms. It highlights the importance of look-ahead masks and the interaction between the first and second multi-headed attention layers, where encoder outputs serve as queries and keys. Finally, it mentions the use of a linear classifier with softmax for generating output probabilities.
- Author
- Afsheen hamid
- Language
- EN