Commits · 67435b574663a25b49263007e7e3b424cd7dd88c · Franck Dary / RL-Parsing · GitLab

Snippets Groups Projects

Jun 25, 2021
- States working for oracle learning · 67435b57
  Franck Dary authored 3 years ago
  
  67435b57
Jun 24, 2021
- Started to work on states · bbdc365a
  Franck Dary authored 3 years ago
  
  bbdc365a
- Corrected bug where predicted columns were not automaticaly set during decoding · 0d5dffb6
  Franck Dary authored 3 years ago
  
  0d5dffb6
Jun 17, 2021
- New parameters : Nb of unaplayable actions picked before breaking the analysis · ead830cc
  Maxime Petit authored 3 years ago
  
  ead830cc
Jun 11, 2021
- Scores are now printed in deterministic order · 85a275f0
  Franck Dary authored 3 years ago
  
  85a275f0
- Corrected a mistake where suffix feature was in fact the prefix. Added the real suffix feature · 085d7252
  Franck Dary authored 3 years ago
  
  085d7252
Jun 10, 2021
- Added suffix feature to neural network · cdfa91d8
  Franck Dary authored 3 years ago
  
  cdfa91d8
Jun 08, 2021
- Added TAG transition and tagparser transition set · fc820be0
  Franck Dary authored 3 years ago
  
  fc820be0
Jun 01, 2021
- Added new reward func · 36342444
  maxime.petit authored 3 years ago
  
  36342444
May 26, 2021
- Fixed reward funcs · 67b8b07c
  maxime.petit authored 3 years ago
  
  67b8b07c
- Added arc swift transition system · 7cc3e7f1
  Franck Dary authored 3 years ago
  
  7cc3e7f1
May 24, 2021
- Added missing argument · 0fa24f11
  Franck Dary authored 3 years ago
  
  0fa24f11
May 19, 2021
- Added LSTM network · 47213de3
  Franck Dary authored 3 years ago
  
  47213de3
May 12, 2021
- Added hyperparameters as program arguments · 2038e1bc
  Franck Dary authored 3 years ago
  
  2038e1bc
- Added multiple reward functions and added program argument to chose them · e8b9c9f0
  Franck Dary authored 3 years ago
  
  e8b9c9f0
- Corrected typos · 2d7c2ee0
  Maxime Petit authored 3 years ago
  
  2d7c2ee0
May 06, 2021
- Print TransitionSet in main · 8cc02de5
  Franck Dary authored 3 years ago
  
  8cc02de5
- Added venv to gitignore · f6be778a
  Franck Dary authored 3 years ago
  
  f6be778a
- Allow several consecutive backs · 72c7e34e
  Franck Dary authored 3 years ago
  
  72c7e34e
- Added incremental mode (-i) : parser cannot see right context if it has not... · 62b38a8c
  Franck Dary authored 3 years ago
  
  Added incremental mode (-i) : parser cannot see right context if it has not been fixated (so only accessible after a BACK)
  62b38a8c
May 04, 2021
- Added French KFold and scripts to train and decode it · d06eac84
  Franck Dary authored 3 years ago
  
  d06eac84
- Corrected path of train log file · fdfc07d3
  Franck Dary authored 3 years ago
  
  fdfc07d3
- Added program argument --ts to specify back actions · dacc99a7
  Franck Dary authored 3 years ago
  
  dacc99a7
- Fixed missing argument · 157585cf
  Franck Dary authored 3 years ago
  
  157585cf
- Improved debug print · 22de30ac
  Franck Dary authored 3 years ago
  
  22de30ac
- modified reward for back · 394637aa
  Maxime Petit authored 3 years ago
  
  394637aa
Apr 28, 2021
- Added history of transitions in feature function · b699477b
  Franck Dary authored 3 years ago
  
  b699477b
- Corrected BACK action : 1) cannot create cycle 2) correct backward movement · 1a127719
  Franck Dary authored 3 years ago
  
  1a127719
Apr 27, 2021
- Added Back transitions, problem : system can cycle · bf43cbf5
  Franck Dary authored 3 years ago
  
  bf43cbf5
- Added history in config · 39e488e9
  Franck Dary authored 3 years ago
  
  39e488e9
- ProbaOracle and probaRandom get smaller with number of epochs · fb6f6ffa
  Franck Dary authored 3 years ago
  
  fb6f6ffa
Apr 26, 2021
- Added form in feature function, and added threshold for number of occurences in dicts · 7729907f
  Franck Dary authored 3 years ago
  
  7729907f
- Refactored feature extraction, and added extractFeature function to NeuralNetwork · 3d80804a
  Franck Dary authored 3 years ago
  
  3d80804a
Apr 24, 2021
- Added bootstrap mode for oracle training · b1c976ae
  Franck Dary authored 3 years ago
  
  b1c976ae
- Reduced probaRandom and probaORacle · e0926705
  Franck Dary authored 3 years ago
  
  e0926705
- In RL, consider non appliable action · 9547a7de
  Franck Dary authored 3 years ago
  
  9547a7de
- Improved debug prints · d36e1f08
  Franck Dary authored 3 years ago
  
  d36e1f08
- Added CUDA support and changed Neural network weight init to avoid Q value explosion in RL · 7ff12b50
  Franck Dary authored 3 years ago
  
  7ff12b50
- Added strucutre containing predicted dependents in Config · bc57dbba
  Franck Dary authored 3 years ago
  
  bc57dbba
- Added random seed as program argument · 436a97bc
  Franck Dary authored 3 years ago
  
  436a97bc