bojone/attention

Issues

• 你好，这个注意力有没有使用的例子呢？如何调用注意力层呢？

  #21 opened 4 years ago by Lerry123
  1

• a = K.permute_dimensions(a, (0, 3, 2, 1))

  #19 opened 5 years ago by monkeyshichi
  2

• Keras的batch_dot()如何用pytorch改写

  #20 opened 5 years ago by linjingxu
  2

• Is there tensorflow implementation of sparse attention?

  #18 opened 5 years ago by hugddygff
  3

• 模型加载问题

  #16 opened 5 years ago by xiong3134
  2

• 请问Q,K,V分别是从哪里输入的呢？

  #14 opened 6 years ago by Ironeie
  2

• K.permute_dimensions(A, (0,3,2,1))运行时报错

  #3 opened 6 years ago by njuccq
  13

• I have no idea what method "to_mask" helps.

  #17 opened 5 years ago by fatLime
  1

• K，Q，V映射

  #15 opened 6 years ago by Vipning
  0

• 不支持mask

  #7 opened 6 years ago by Zhao6924
  2

• attention_keras.py中，Position_Embedding的pos的计算

  #10 opened 6 years ago by rivershadowy
  2

• attention_keras.py

  #13 opened 6 years ago by myndtt
  0

• 请问一下，tf版的78，80行的转置操作是什么意思呢？ 去掉这两行进行训练对结果也没什么影响。

  #12 opened 6 years ago by yangjun023
  2

• tf中position

  #11 opened 6 years ago by lrx1213
  4

• position_ij = tf.concat([tf.cos(position_ij), tf.sin(position_ij)], 1)

  #5 opened 6 years ago by imhuay
  1

• 感觉好像少了参数Wo

  #8 opened 6 years ago by duanyu
  1

• 行为识别　用attention

  #9 opened 6 years ago by guofuzheng
  1

• How to understand the decoder input(target)?

  #6 opened 6 years ago by zoe218
  1

• position_ij = tf.matmul(position_i, position_j) 应该用 multiply 而不是 matmul

  #4 opened 7 years ago by imhuay
  1

• Position_Embedding肯能有些问题

  #2 opened 7 years ago by liangoy
  1

• 请问 Position Embedding 可以被训练吗？

  #1 opened 7 years ago by ZJUguquan
  3