brightmart/bert_language_understanding

Issues

• 你的textcnn实验，只要把学习率调大一个数量级，没有预训练也可以得到一样的结果

  #18 opened 5 years ago by wykdg
  5

• 运行run_classifier_predict_online.py出错

  #25 opened 5 years ago by Vincent131499
  1

• using bert with unused/discarded data

  #24 opened 5 years ago by Priyansh2
  0

• run_classifier_predict_online.py

  #16 opened 6 years ago by bihui9968
  4

• error in run_classifier_predict_online.py

  #22 opened 5 years ago by xikunlun001
  1

• about pre_train

  #21 opened 5 years ago by charlesfufu
  0

• 关于预训练模型设计

  #19 opened 5 years ago by yun97
  0

• 关于执行pre_train命令的问题

  #17 opened 6 years ago by c0derm4n
  0

• bug

  #15 opened 6 years ago by bihui9968
  0

• What is the accuracy of mask word prediction?

  #14 opened 6 years ago by guotong1988
  1

• different result when set lr to 0.001

  #13 opened 6 years ago by liu-nlper
  4

• About "scaled_dot_product_attention_batch"

  #9 opened 6 years ago by yuanxiaosc
  1

• tokenize_style=char的问题

  #10 opened 6 years ago by godfatherzzx
  1

• The reason why pretrain does work.

  #12 opened 6 years ago by guotong1988
  1

• question about learning rate

  #11 opened 6 years ago by liu-nlper
  2

• Bert_model doesn't work

  #7 opened 6 years ago by geogreff
  2

• the pre-trained MLM performance

  #6 opened 6 years ago by yyht
  2

• question on the implement of bert_cnn_model

  #4 opened 6 years ago by zhhongzhi
  5

• pre-trained word embedding的问题

  #5 opened 6 years ago by ethereal666
  1

• What is the final f1 performance?

  #3 opened 6 years ago by guotong1988
  2

• Three undefined names

  #2 opened 6 years ago by cclauss
  3