Commits · 5b18cb1d8f1bc825ec1c3d427ab5889b77084cda · shensq / mi_counseling

Mar 27, 2020
- add full support for k-bert scheme · 5b18cb1d
  shensq authored 4 years ago
  
  5b18cb1d
Mar 13, 2020
- fix the split to 90-5-5 · 3f4090ec
  shensq authored 5 years ago
  
  3f4090ec
- without kbert now also works · 763bf8e5
  shensq authored 5 years ago
  
  763bf8e5
- output_attention mask can use gpu · aa653442
  shensq authored 5 years ago
  
  aa653442
- fix a bug that the output can not be longer than ground truth · bf8bdb91
  shensq authored 5 years ago
  
  bf8bdb91
- the mask in sample is now working · 139c8d03
  shensq authored 5 years ago
  
  139c8d03
- debuging sample · 892a8c4a
  shensq authored 5 years ago
  
  892a8c4a
- debuging sample · 6a64f2ae
  shensq authored 5 years ago
  
  6a64f2ae
- attention_mask move to gpu · 9dda1ef1
  shensq authored 5 years ago
  
  9dda1ef1
- attention mask in decoding · 95f573b4
  shensq authored 5 years ago
  
  95f573b4
- run with mask in the sh file · 3470baf0
  shensq authored 5 years ago
  
  3470baf0
- sample now support attention mask · ee3247e9
  shensq authored 5 years ago
  
  ee3247e9
- add kbert masking · 8d89aedc
  shensq authored 5 years ago
  
  8d89aedc
Mar 10, 2020
- change the data split to 90-5-5 · d853a709
  shensq authored 5 years ago
  
  d853a709
- use early stop · 0aa2435a
  shensq authored 5 years ago
  
  0aa2435a
- fix a but that only one clause will be load for a location · c169f46a
  shensq authored 5 years ago
  
  c169f46a
- Gptdataset_full now use x_y_with_comet · 0a58afd0
  shensq authored 5 years ago
  
  0a58afd0
- sampling now compatible with kbert embedding · 2eaec2db
  shensq authored 5 years ago
  
  2eaec2db
- Add kbert soft-position. Attention mask is not implemented · f3f3048d
  shensq authored 5 years ago
  
  f3f3048d
Feb 27, 2020
- Updated · 98e0ca30
  DeepLearning VM authored 5 years ago
  
  98e0ca30