Commits · db63d7971b649d13b57d8403ebe9fc841e132993 · Sciencebeam / Sciencebeam Gym

Jan 02, 2018
- optionally use specific temp bucket when running ml-engine jobs · dbdb16b8
  Daniel Ecer authored 7 years ago
  
  dbdb16b8
Dec 18, 2017
- don't store job id to a file anymore · fa1099cb
  Daniel Ecer authored 7 years ago
  
  fa1099cb
- add separate job prefix for train, save and predict · 744ba618
  Daniel Ecer authored 7 years ago
  
  744ba618
Dec 08, 2017
- increased google ml runtime environment to 1.2 · e7c3e449
  Daniel Ecer authored 7 years ago
  
  e7c3e449
Dec 07, 2017
- added MAX_TRAIN_STEPS script configuration; fixed passing in CLASS_WEIGHTS_FILENAME · 6d7a715d
  Daniel Ecer authored 7 years ago
  
  6d7a715d
- train with class weights · c5f49894
  Daniel Ecer authored 7 years ago
  
  c5f49894
Dec 04, 2017
- use new channel map format; allow channel names to be specified · 17415266
  Daniel Ecer authored 7 years ago
  
  17415266
Dec 01, 2017
- using new pipeline and amended config · 9e377a60
  Daniel Ecer authored 7 years ago
  
  9e377a60
Aug 03, 2017
- minor quantitative should have been qualitative in this case · 9faae083
  Daniel Ecer authored 7 years ago
  
  9faae083
- make batch size configurable · ba0ae9c5
  Daniel Ecer authored 7 years ago
  
  ba0ae9c5
- added optional cross entropy loss · ebe227b2
  Daniel Ecer authored 7 years ago
  
  ebe227b2
Aug 02, 2017
- setting random seed · 05f7d201
  Daniel Ecer authored 7 years ago
  
  05f7d201
Aug 01, 2017
- added explicit quantitative evaluation · fd816b46
  Daniel Ecer authored 7 years ago
  
  fd816b46
- added outputs_most_likely for RGB mode w/ scores · 36166239
  Daniel Ecer authored 7 years ago
  
  36166239
Jul 28, 2017
- added missing $TRAINING_ARGS · da9e4a89
  Daniel Ecer authored 7 years ago
  
  da9e4a89
Jul 19, 2017
- initial commit · 32fbd1c1
  Daniel Ecer authored 7 years ago
  
  32fbd1c1