Show, Attend and Tell: Soft Attention Context
Beginner
The problem, from first principles
Show, Attend and Tell (Xu et al., 2015) captions images by attending to regions of a CNN feature map at each word. The context for a word is the attention-weighted average of the region features.
From theory to code
Implement soft_context(alpha, feats), the attention-weighted sum of the image features.
Constraints
- Weights sum to one.
Hints
Hint 1
Multiply the weight vector by the feature matrix.
Keywords
research-paperssequence-modelsattentioncaptioning
Python
Run your code to see results
No results yet
Run your code to check it against the examples, or submit to run every test.