题目
题目

CS-7643-O01, OAN, OSZ Quiz #4: Module 3

多项填空题

As defined in Attention is All You Need, what is the size of the cross-attention matrix between the encoder and decoder given the following English to Spanish translation: I am very handsome -> Soy muy guapo Please assume the following: d_k = d_q = 64 d_v = 32 Please ignore the <SOS> and <EOS> tokens. cross-attention means Attention(Q, K, V) NOTE: Please round to the nearest integer. [Fill in the blank] rows[Fill in the blank] columns

查看解析

查看解析

标准答案
Please login to view
思路分析
To determine the size of the cross-attention matrix, we need to recall how cross-attention is defined in Transformer architectures. The cross-attention matrix is formed by Attention(Q, K, V), where Q comes from the decoder (the current target sequence being generated) and K and V come from the encoder (the source sequence). The dimensions of this mat......Login to view full explanation

登录即可查看完整答案

我们收录了全球超50000道考试原题与详细解析,现在登录,立即获得答案。

类似问题

  On scaled dot-product attention and training stability of a transformer:   I Without scaling by 𝐷 𝑘 , the variance of the dot product 𝑞 𝑛 ⊤ 𝑘 𝑚 grows with dimensionality, producing large logits that can saturate the softmax. II Scaling by 𝐷 𝑘 primarily solves exploding-gradient problems inside the value projection 𝑉 . III The softmax-normalized matrix S o f t m a x ( 𝑄 𝐾 ⊤ ) is applied row-wise, making each row represent how strongly a query attends to all keys. IV Scaled dot-product attention computes A t t e n t i o n ( 𝑄 , 𝐾 , 𝑉 ) = S o f t m a x ! ( 𝑄 𝐾 ⊤ 𝐷 𝑘 ) 𝑉 , and the resulting matrix always has the same dimension as 𝑉 .  

Which innovation is at the core of the transformer architecture and enables modeling long-range dependencies effectively?

Which of the following attention models uses a subset of the input to derive the output, and can not be trained directly with gradient methods?

Why is the attention mechanism particularly suitable for modeling financial time series? 

更多留学生实用工具

加入我们，立即解锁 海量真题 与 独家解析，让复习快人一步！