ciprian-chelba/1-billion-word-language-modeling-benchmark

Issues

• Data Sources for the Corpus

  #7 opened 2 years ago by nadyadtm
  2

• if the word not in vocab， what should I do？ or it always can't happen because the FullTokenizer

  #6 opened 6 years ago by wangwang110
  1

• Some Training Data Duplicated in Heldout Data

  #5 opened 6 years ago by bjascob
  10

• Dev / Test set?

  #4 opened 7 years ago by eric-haibin-lin
  4

• Dead code.google.com link

  #3 opened 7 years ago by charlesreid1
  1

• meaning of cost values in output.tar

  #2 opened 8 years ago by jowagner
  2

• question on the corpus size / script

  #1 opened 8 years ago by vince62s
  2