pbloem/former

Issues

• No Field in torchtext.data

  #35 opened a year ago by dtdo90
  1

• Question about k × hk weight matrices

  #30 opened a year ago by ramesaliyev
  1

• Why the use of log_softmax ?

  #36 opened a year ago by dsantiago
  1

• How to tokenize a testing phrase

  #37 opened a year ago by edgarmg91
  0

• ModuleNotFoundError: No module named 'past'

  #1 opened 4 years ago by koen-dejonghe
  7

• Calculation of memory required in "Going big"

  #34 opened a year ago by oleksandrasaskia
  2

• Einsum to avoid transpose and reshape

  #4 opened 5 years ago by azzever
  3

• Question about the tool used to make the figures in the blog post

  #11 opened 5 years ago by yassouali
  2

• improve performance in deeper network using your multi head attention code

  #29 opened 2 years ago by fatemehniknezhad
  3

• Hi,question about sliced-up version self-attention

  #28 opened 3 years ago by Faith-Uchiha
  1

• Module import error

  #25 opened 3 years ago by Cybernetic1
  1

• conda installation failed

  #27 opened 3 years ago by bsun0802
  1

• Understanding multiheaded attention.

  #23 opened 3 years ago by anorak94
  0

• Blog is down

  #26 opened 3 years ago by JJGO
  1

• ImportError: attempted relative import with no known parent package

  #24 opened 3 years ago by RubenRidderstrom
  8

• data.Field no longer supported in torchtext

  #20 opened 3 years ago by jvdburgt
  1

• ModuleNotFoundError: No module named 'past'

  #9 opened 4 years ago by jjong2ya
  5

• High compute time, what is a reasonable generator model to get somewhat good results to play with?

  #22 opened 4 years ago by jplasser
  1

• Is narrow attention implemented correctly?

  #13 opened 4 years ago by TheGrayFrost
  8

• slide 50, v and q mixed up

  #19 opened 4 years ago by ulf1
  1

• token_embedding for non-text sequences

  #18 opened 4 years ago by StolkArjen
  6

• Weight matrix dimensions after transofrmation

  #17 opened 4 years ago by SrinjaySarkar
  1

• Using the trained model

  #14 opened 4 years ago by ShivanshuPurohit
  1

• How can we visualize the self-attention map

  #15 opened 4 years ago by amiltonwong
  1

• Why is dividing by e**(1/4) for both keys and queries more memory efficient?

  #5 opened 5 years ago by esvhd
  4

• Accuracies on the examples

  #12 opened 5 years ago by sf-wind
  1

• Issue with masking

  #10 opened 5 years ago by mc-robinson
  1

• Comparing SelfAttention classes

  #8 opened 5 years ago by sidneyaraujomelo
  2

• AttributeError: module 'torch' has no attribute 'triu_indices'

  #7 opened 5 years ago by xxxxyan
  5

• Masking done for the upper or lower triangle?

  #6 opened 5 years ago by esvhd
  1

• Weight scaling should be for keys not values?

  #2 opened 5 years ago by esvhd
  2