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 |
---|---|---|
Retrieval | 93 | 10.84% |
Language Modelling | 70 | 8.16% |
Question Answering | 51 | 5.94% |
Large Language Model | 47 | 5.48% |
Text Generation | 23 | 2.68% |
Sentence | 21 | 2.45% |
In-Context Learning | 19 | 2.21% |
Code Generation | 17 | 1.98% |
Information Retrieval | 17 | 1.98% |