NormXU/ERNIE-Layout-Pytorch

Issues

• 有对应的使用torch训练ernire-layout的代码？

  #23 opened 3 months ago by tianchiguaixia
  0

• Which version of Flash Attention has been used in this project?

  #22 opened 3 months ago by 14H034160212
  2

• Using the new extended seq length

  #20 opened 9 months ago by rbvh
  5

• A question about tokenizer

  #21 opened a year ago by Jermaine1996
  2

• long input - larger than 510 tokens

  #17 opened a year ago by DiddyC
  2

• Tokenizer does not support boxes as input

  #14 opened a year ago by logan-markewich
  4

• Can convert2torch.py convert fine tined models?

  #19 opened 2 years ago by htcml
  2

• misleading between num_classes and num_labels

  #18 opened 2 years ago by huyhoang17
  1

• not initialized warning for some weights

  #16 opened 2 years ago by DiddyC
  5

• Should we use 2 points or point + width/height in bbox for model training?

  #15 opened 2 years ago by calvinzhan
  1

• Couldn't get good result with actual image

  #13 opened 2 years ago by calvinzhan
  1

• DocVQA fine-tuning example?

  #9 opened 2 years ago by JP-Leite
  1

• License of this repo

  #11 opened 2 years ago by sandorkonya
  1

• List index out of range

  #1 opened 2 years ago by karndeb
  2

• different label pad token id between layoutxlm and ernie layout

  #8 opened 2 years ago by vanpersie32
  1

• Tokenizer for ernie-layoutx-base-uncased

  #5 opened 2 years ago by maxjeblick
  8

• Ernie-Layout english weights?

  #7 opened 2 years ago by logan-markewich
  2

• test_ernie_token_cls gives UnboundLocalError: local variable 'eop_idx' referenced before assignment

  #6 opened 2 years ago by maxjeblick
  5

• tokenizer can not return offset_map

  #4 opened 2 years ago by WallE-Chang
  2

• diff between paddle and torch

  #3 opened 2 years ago by WallE-Chang
  7

• Could you make test_ernie_qa runnable?

  #2 opened 2 years ago by calvinzhan
  1