Attention Dropout is a type of dropout used in attention-based architectures, where elements are randomly dropped out of the softmax in the attention equation. For example, for scaled-dot product attention, we would drop elements from the first term:
$$ {\text{Attention}}(Q, K, V) = \text{softmax}\left(\frac{QK^{T}}{\sqrt{d_k}}\right)V $$
Paper | Code | Results | Date | Stars |
---|
Task | Papers | Share |
---|---|---|
Language Modelling | 82 | 9.80% |
Retrieval | 76 | 9.08% |
Question Answering | 48 | 5.73% |
Large Language Model | 43 | 5.14% |
Sentence | 29 | 3.46% |
In-Context Learning | 22 | 2.63% |
Text Generation | 21 | 2.51% |
Information Retrieval | 19 | 2.27% |
Code Generation | 14 | 1.67% |