Commits · e9c362bea4f271f298853c2a894e67231825c269 · Patrick Skillman-Lawrence / mhc_rank

Nov 11, 2019
- In train_allele_specific_models_command.py, serial run is when num jobs = 0 (not 1). · e9c362be
  Tim O'Donnell authored 5 years ago
  
  Fixes #149
  e9c362be
Sep 05, 2019
- pylint nits · 7f1f671d
  Tim O'Donnell authored 5 years ago
  
  7f1f671d
- Docs and unit test fixes · f266022a
  Tim O'Donnell authored 5 years ago
  
  f266022a
Sep 04, 2019
- cluster parallelism · f56a579c
  Tim O'Donnell authored 5 years ago
  
  f56a579c
- fix · 32d0bd8e
  Tim O'Donnell authored 5 years ago
  
  32d0bd8e
Dec 22, 2018
- First cut on Kim 2014 benchmark script · 3dd8150e
  Tim O'Donnell authored 6 years ago
  
  3dd8150e
Feb 20, 2018
- attempted travis fix · 7cbdc4d9
  Tim O'Donnell authored 7 years ago
  
  7cbdc4d9
- fix · 75dec823
  Tim O'Donnell authored 7 years ago
  
  75dec823
- remove obsolete downloads · 42ba21de
  Tim O'Donnell authored 7 years ago
  
  42ba21de
- Hold out 10% of data in models_class_unselected · 548ad025
  Tim O'Donnell authored 7 years ago
  
  548ad025
- factor out parallelism related argparse arguments · aa05cea9
  Tim O'Donnell authored 7 years ago
  
  aa05cea9
Feb 19, 2018
- Basic support for model selection · 11ad19cd
  Tim O'Donnell authored 7 years ago
  
  11ad19cd
- py 2.7 fix · 1c931bc6
  Tim O'Donnell authored 7 years ago
  
  1c931bc6
Feb 15, 2018
- fixes · 402d3176
  Tim O'Donnell authored 7 years ago
  
  402d3176
Feb 14, 2018
- fix · b3be67c6
  Tim O'Donnell authored 7 years ago
  
  b3be67c6
- fix · 0dc5caff
  Tim O'Donnell authored 7 years ago
  
  0dc5caff
- fix · 9f1bb326
  Tim O'Donnell authored 7 years ago
  
  9f1bb326
- first cut on multi-round training · 51dabf24
  Tim O'Donnell authored 7 years ago
  
  51dabf24
Feb 13, 2018
- more parallelism · f4d66885
  Tim O'Donnell authored 7 years ago
  
  f4d66885
- Reseed the numpy random number generator in each worker · 31751035
  Tim O'Donnell authored 7 years ago
  
  31751035
Feb 11, 2018
- Add parallelism.py. Expand hyperparameter search." · b9fd472a
  Tim O'Donnell authored 7 years ago
  
  b9fd472a
- better handling of crashed workers · 4752df65
  Tim O'Donnell authored 7 years ago
  
  4752df65
- Restart workers every N tasks · cc1543cf
  Tim O'Donnell authored 7 years ago
  
  cc1543cf
Feb 10, 2018
- fix · 51573830
  Tim O'Donnell authored 7 years ago
  
  51573830
- Continually save models as they are trained, even when running in parallel · 220aaf31
  Tim O'Donnell authored 7 years ago
  
  220aaf31
Feb 09, 2018
- fix · bbba5e7d
  Tim O'Donnell authored 7 years ago
  
  bbba5e7d
- Include training data subset as a hyperparameter · 505899a5
  Tim O'Donnell authored 7 years ago
  
  505899a5
- Fix · 8b9e468f
  Tim O'Donnell authored 7 years ago
  
  8b9e468f
- fixes · 03dcdb51
  Tim O'Donnell authored 7 years ago
  
  03dcdb51
- better gpu/cpu allocation · a9ec7e1c
  Tim O'Donnell authored 7 years ago
  
  a9ec7e1c
- mixed gpu / cpu training hack · 4ea2b240
  Tim O'Donnell authored 7 years ago
  
  4ea2b240
- update comments · f472d4b7
  Tim O'Donnell authored 7 years ago
  
  f472d4b7
- Multi-GPU training · c0479deb
  Tim O'Donnell authored 7 years ago
  
  c0479deb
Feb 08, 2018
- Workaround tqdm issue · 877c60a0
  Tim O'Donnell authored 7 years ago
  
  Add suggested workaround for https://github.com/tqdm/tqdm/issues/481
  877c60a0
Feb 06, 2018
- lazy-load model weights · 33e8c10a
  Tim O'Donnell authored 7 years ago
  
  33e8c10a
- fix · 6dcc5a8c
  Tim O'Donnell authored 7 years ago
  
  6dcc5a8c
- specify parallelization for train and calibrate percentile ranks seperately · 038704d2
  Tim O'Donnell authored 7 years ago
  
  038704d2
- improve parallelization · 77fda00d
  Tim O'Donnell authored 7 years ago
  
  77fda00d
- fix · fc83dd0b
  Tim O'Donnell authored 7 years ago
  
  fc83dd0b
- attempt fix deadlock · 01affd07
  Tim O'Donnell authored 7 years ago
  
  01affd07