jadore801120/attention-is-all-you-need-pytorch

Issues

• python preprocess.py -lang_src de -lang_trg en -save_data multi30k_de_en.pkl -share_vocab have problem

  #222 opened 5 months ago by yyydfff
  1

• 根据requirements.txt安装出错

  #215 opened a year ago by kmphuang
  4

• when I run the `python preprocess.py -lang_src de -lang_trg en -share_vocab -save_data m30k_deen_shr.pkl`.I have faced a problem

  #218 opened 10 months ago by dapaolufuduizhang
  1

• Error time to execute 'python preprocess.py -lang_src de -lang_trg en -share_vocab -save_data m30k_deen_shr.pkl' command

  #216 opened a year ago by evilczy
  3

• preprocess error

  #202 opened 2 years ago by zhoup150344
  6

• About target mask

  #217 opened a year ago by KimRass
  0

• ValueError: Cell is empty

  #204 opened 2 years ago by Kznnd
  4

• TranslationDataset is now deprecated in torchtext

  #194 opened 3 years ago by imkzh
  2

• download dataset error

  #197 opened 3 years ago by qimg412
  4

• why is masking performed again during the inference decoder stage?

  #213 opened a year ago by Akshay1-6180
  0

• Possible mistakes in d_k, d_v of MultiheadAttention

  #211 opened a year ago by SARIHUST
  0

• In patch_trg, i cant understand why do you change the data shape like that

  #205 opened 2 years ago by kwanhoP
  3

• May I ask you a question about the "scale_emb_or_prj" parameter?

  #210 opened 2 years ago by aitch25
  0

• Performance Confusion

  #209 opened 2 years ago by Zarca
  0

• Is there some way to convert .chkpt to other form model such as onnx?

  #208 opened 2 years ago by warren-wzw
  0

• The results of the translate function.

  #176 opened 4 years ago by zshyang
  4

• How to export the trained chkpt network to onnx?

  #172 opened 4 years ago by ZhangDongyuCN
  5

• Attention value is strange

  #182 opened 3 years ago by YPatrickW
  1

• CUBLAS_STATUS_EXECUTION_FAILED when calling `cublasSgemm( handle, opa, opb, m, n, k, &alpha, a, lda, b, ldb, &beta, c, ldc)`

  #193 opened 3 years ago by lisp2047
  1

• why dropping last example with patch_src&patch_trg function @train.py

  #174 opened 4 years ago by pluspluswu
  1

• some problem

  #185 opened 3 years ago by zhLia
  3

• what's the intuition behind getting q, k and v from embedding

  #201 opened 2 years ago by ShouravBR
  1

• My question

  #195 opened 3 years ago by Messiz
  2

• OverflowError

  #198 opened 3 years ago by Daming-TF
  0

• Can't find model 'en'

  #169 opened 4 years ago by manhph2211
  2

• Confusion regarding embedding space

  #186 opened 3 years ago by IamAdiSri
  2

• Beam search torch.log

  #191 opened 3 years ago by actforjason
  0

• test error

  #181 opened 3 years ago by LonelyPlanetIoT
  2

• 变量名能不能起得好读一点？

  #183 opened 3 years ago by Xelawk
  0

• learning rate update before optimizer.step()

  #188 opened 3 years ago by AlbertiPot
  1

• why PEpos+k can be represented as a linear function of PEpos?

  #187 opened 3 years ago by myrainbowandsky
  0

• Resuming Training

  #158 opened 4 years ago by kaiyon07
  5

• MultiHeadAttention input shape

  #179 opened 4 years ago by Superklez
  0

• Incorrect implementation?

  #177 opened 4 years ago by weilueluo
  2

• Input sequence dimensions of MultiHeadAttention

  #178 opened 4 years ago by Superklez
  1

• If we change the code in Model.py like this, the convergence speed would be faster.

  #175 opened 4 years ago by Sry2016
  1

• Why the previous version train faster

  #157 opened 4 years ago by dwtenis
  1

• How to deal with the UNK_TOKEN?

  #173 opened 4 years ago by lhy2749
  0

• why none pad mask is nedd

  #166 opened 4 years ago by helloworld729
  1

• train big data(8G)

  #171 opened 4 years ago by JoeCoding
  0

• raise ConnectionError(e, request=request)

  #156 opened 4 years ago by KrisLee512
  1

• Question About Attention Score Computation Process & Intuition

  #167 opened 4 years ago by rezhv
  0

• what is meaning of trg_pad_idx in label smoothing loss?

  #165 opened 4 years ago by fakerhbj
  0

• wrong with the code!!!!!

  #164 opened 4 years ago by chenrxi
  0

• what does n_head, d_model, d_k, d_v stands for?

  #162 opened 4 years ago by seyeeet
  1

• SyntaxError: invalid syntax

  #163 opened 4 years ago by junzew
  1

• Why decoding is needed during inference ?

  #160 opened 4 years ago by rajeevbaalwan
  0

• How does the gradients flow in cal_loss function in train.py?

  #159 opened 4 years ago by InhyeokYoo
  0

• Surprising PPL on WMT 17

  #154 opened 4 years ago by luffycodes
  0

• d_k not equal to d_k gives issues

  #153 opened 5 years ago by luffycodes
  0