Commits · 028cfe6162da7332aa567fcdc5ce99cbf1c07edf · mirrored_repos / MachineLearning / thukeg / SwissArmyTransformer

This project is mirrored from https://github.com/THUDM/SwissArmyTransformer. Pull mirroring failed 2 months ago.
Repository mirroring has been paused due to too many failed attempts. It can be resumed by a project maintainer or owner.
Last successful update 2 months ago.

Oct 31, 2021
- Fix get_params_for_weight_decay_optimization · 028cfe61
  Zhengxiao Du authored 3 years ago
  
  028cfe61
- support beam search · 69cbe5dd
  Ming Ding authored 3 years ago
  
  69cbe5dd
Oct 30, 2021
- fix bugs & pass cogview2 and glm · 26029fec
  Ming Ding authored 3 years ago
  
  26029fec
Oct 29, 2021
- pass cogview generate · 78b7ca5f
  Ming Ding authored 3 years ago
  
  78b7ca5f
- package prepare · cb3af696
  Ming Ding authored 3 years ago
  
  cb3af696
- Merge remote-tracking branch 'origin/dev' into dev · a61c570f
  Ming Ding authored 3 years ago
  
  a61c570f
- Merge remote-tracking branch 'origin/fix_valid_bug' into dev · a374aeb7
  Ming Ding authored 3 years ago
  
  a374aeb7
Oct 28, 2021
- Add Chinese GLM model · 04752ae3
  Zhengxiao Du authored 3 years ago
  
  04752ae3
- fix multiple fill back bug · 2fa0908e
  Ming Ding authored 3 years ago
  
  2fa0908e
- fix glm mask bug · 3d55652b
  Ming Ding authored 3 years ago
  
  3d55652b
Oct 27, 2021
- tmp save dev to dzx review glm · f8d2632a
  Ming Ding authored 3 years ago
  
  f8d2632a
- solve conflit · 031c3f3c
  zhuoyiyang authored 3 years ago
  
  031c3f3c
- fix out range bug · a0a295d2
  zhuoyiyang authored 3 years ago
  
  a0a295d2
- change to RandomMapping · 16378f14
  zhuoyiyang authored 3 years ago
  
  16378f14
Oct 26, 2021
- fix valid dataset bug · a36b9ca3
  zhuoyiyang authored 3 years ago
  
  a36b9ca3
Oct 24, 2021
- merge glm · ebd91ef0
  Ming Ding authored 3 years ago
  
  ebd91ef0
Oct 23, 2021
- Implement no repeat ngram and min target length strategies · 2515985c
  Zhengxiao Du authored 3 years ago
  
  2515985c
- Fix beam search strategy · d59e2a62
  Zhengxiao Du authored 3 years ago
  
  d59e2a62
Oct 22, 2021
- fix device inconsistent bug & add forward hooks · 465c146b
  Ming Ding authored 3 years ago
  
  465c146b
- Implement beam search strategy · 94c3c3eb
  Zhengxiao Du authored 3 years ago
  
  94c3c3eb
- Fix attention mask · e9ae3132
  Zhengxiao Du authored 3 years ago
  
  e9ae3132
- Merge branch 'main' into glm · 7c35e218
  Zhengxiao Du authored 3 years ago
  
  7c35e218
- Fix GLM inference · d9a54331
  Zhengxiao Du authored 3 years ago
  
  d9a54331
Oct 21, 2021
- fixed mutiple save bug · 52077b84
  Ming Ding authored 3 years ago
  
  52077b84
Oct 20, 2021
- Fix attention mask in GLM · d039667d
  Zhengxiao Du authored 3 years ago
  
  d039667d
- Import GLM inference code · ba522b5e
  Zhengxiao Du authored 3 years ago
  
  ba522b5e
Oct 19, 2021
- fix loss_checker bug & slice · 3e304e54
  Ming Ding authored 3 years ago
  
  3e304e54
Oct 18, 2021
- Add GLM Model · 9833cc8a
  Zhengxiao Du authored 3 years ago
  
  9833cc8a
Oct 14, 2021
- prepare for large-scale generation · 8584bf91
  Ming Ding authored 3 years ago
  
  8584bf91
Oct 10, 2021
- del old sampling · a2625727
  Ming Ding authored 3 years ago
  
  a2625727
- del some unused things · e3f4751f
  Ming Ding authored 3 years ago
  
  e3f4751f
- finish testing cogview2 generation · b32472a9
  Ming Ding authored 3 years ago
  
  b32472a9
Oct 09, 2021
- test cogview1 generation, pass · 154fc361
  Ming Ding authored 3 years ago
  
  154fc361
Oct 08, 2021
- tmp finish naive ar sampling · 7c5a12da
  Ming Ding authored 3 years ago
  
  7c5a12da
Oct 07, 2021
- test load model, perform normally · 7ea243f0
  Ming Ding authored 3 years ago
  
  7ea243f0
Oct 06, 2021
- Merge branch 'main' of github.com:THUDM/SwissArmyTransformer into main · 99c7a961
  Ming Ding authored 3 years ago
  
  99c7a961
- tokenization · 229f37e1
  Ming Ding authored 3 years ago
  
  229f37e1
- Merge branch 'main' of github.com:THUDM/SwissArmyTransformer into main · 85d98d39
  Ming Ding authored 3 years ago
  
  85d98d39
- tmp save 4 · 142df57c
  Ming Ding authored 3 years ago
  
  142df57c
- del old · 39e6d723
  Ming Ding authored 3 years ago
  
  39e6d723