vrenkens/nabu

Issues

• Performance issue in the definition of marigin_loss, nabu/neuralnetworks/trainers/loss_functions.py(P1)

  #60 opened 3 years ago by DLPerf
  0

• Joint CTC-Attention

  #22 opened 6 years ago
  10

• Multi-task learning

  #59 opened 5 years ago by KunalDhawan
  0

• Test loss = 1 / decoder giving back labels

  #58 opened 6 years ago by jtcampbellsoup
  3

• tensorflow.python.framework.errors_impl.InvalidArgumentError: assertion failed: [Jeff thought you argued in favor of a centrifuge purchase. <eos>]

  #55 opened 6 years ago by nipengmath
  9

• DBLSTM/librispeech recipe

  #47 opened 6 years ago by AzizCode92
  1

• Would you please give an example about the data formet?

  #56 opened 6 years ago by helloWaterM
  0

• CTC decoder results saving

  #51 opened 6 years ago by rzcwade
  4

• data overlapped in itself in tensorboard (validation loss)

  #50 opened 6 years ago by AzizCode92
  3

• Training a language model for rescoring outputs

  #54 opened 6 years ago by k-priyadarshi
  2

• Hi Vincent! Is there any way I can train a language model on my dataset using Nabu for rescoring the outputs? Please reply.

  #53 opened 6 years ago by k-priyadarshi
  1

• Enabling the layer normalization decrease the performance of the model

  #44 opened 6 years ago by AzizCode92
  3

• Interpreting test loss as PER?

  #49 opened 6 years ago by k-priyadarshi
  2

• sync optimizer problem

  #43 opened 6 years ago by fanlu
  15

• windowing attention

  #48 opened 6 years ago by riyijiye
  2

• Clarification required for some methods

  #45 opened 6 years ago by k-priyadarshi
  2

• Beam_search_decoder error of "list index out of range"

  #42 opened 6 years ago by AzizCode92
  7

• _cut_sequence issue

  #38 opened 6 years ago by rzcwade
  4

• Single Machine Mode with Multi-GPU

  #39 opened 6 years ago by AzizCode92
  6

• Name of folder standart should be renamed to standard

  #40 opened 6 years ago by AzizCode92
  1

• phoneme vs. grapheme based?

  #35 opened 6 years ago by mjhanphd
  2

• which recipe to suggest for librispeech

  #30 opened 6 years ago by riyijiye
  5

• your local BeamSearchDecoder implementation

  #28 opened 6 years ago by riyijiye
  3

• the model converge so slow

  #26 opened 6 years ago by betterdam
  2

• Distribute training problem

  #31 opened 7 years ago by fanlu
  7

• pBLSTM not implemented for LAS/WSJ recipe?

  #37 opened 6 years ago by rzcwade
  2

• encoded and encoded_seq_length in attention.py

  #36 opened 7 years ago by rzcwade
  2

• issue with training (LAS TIMIT)

  #32 opened 7 years ago by rzcwade
  12

• kaldi use mfcc feature,but fbank in config/recipes/DNN/WSJ/feature_processor.cfg

  #19 opened 7 years ago by fanlu
  20

• LAS/WSJ

  #33 opened 7 years ago by rzcwade
  4

• Training loss over 400?

  #34 opened 7 years ago by rzcwade
  2

• Invalid Argument Error when training

  #8 opened 7 years ago by jannaescur
  5

• handling <sos> and <eos>

  #29 opened 7 years ago by riyijiye
  5

• speller decoder construction

  #27 opened 7 years ago by riyijiye
  2

• pblstm implementation

  #25 opened 7 years ago by riyijiye
  2

• results tensorboard

  #13 opened 7 years ago by jannaescur
  3

• timit las loss large

  #20 opened 7 years ago by wbgxx333
  1

• question about speller

  #21 opened 7 years ago by riyijiye
  2

• Distributed computing

  #23 opened 7 years ago by riyijiye
  2

• Why do ‘<sos>’ and '<eos>' have the same label?

  #24 opened 7 years ago by ZhengkunTian
  2

• How to calculate the WER for LAS model

  #7 opened 7 years ago by brijmohan
  3

• error while read data when I use recipe DBLSTM

  #18 opened 7 years ago by fanlu
  2

• module can't find

  #17 opened 7 years ago by fanlu
  7

• LAS paper

  #11 opened 7 years ago by jannaescur
  1

• fisher information

  #14 opened 7 years ago by jannaescur
  3

• What is the relationship between this project and tfkaldi?

  #16 opened 7 years ago by fanlu
  2

• Character Error rate (CER) or Phoneme Error Rate (PER) ?

  #15 opened 7 years ago by xavigiro
  2

• evaluate results

  #12 opened 7 years ago by jannaescur
  14

• symbolic link 'model' directory

  #10 opened 7 years ago by jannaescur
  7

• Label value outside the valid range

  #9 opened 7 years ago by jannaescur
  1