Commits · 22975fa7b1b670ebd5de9b6ddc754a65c70c5140 · OpenDAS / torchani · GitLab

13 Aug, 2018 3 commits
- fix padding strip when batching (#65) · 22975fa7
  Gao, Xiang authored Aug 13, 2018
  
  22975fa7
- preprocess data on CPU to save cuda memory (#64) · 7c9402d0
  Gao, Xiang authored Aug 13, 2018
  
  7c9402d0
- Modify training related APIs to batch the whole data by padding (#63) · 50fb6cdb
  Gao, Xiang authored Aug 13, 2018
  
  50fb6cdb
11 Aug, 2018 1 commit
- Support batching non isomers in models and aev computer (#62) · b7cab4f1
  Gao, Xiang authored Aug 11, 2018
  
  b7cab4f1
10 Aug, 2018 2 commits
- add TransformedLoss to allow adding non-linearity to computed loss (#61) · b63f6e40
  Gao, Xiang authored Aug 10, 2018
  
  b63f6e40
- use more specific ops for better performance (#60) · b16a3234
  Gao, Xiang authored Aug 09, 2018
  
  b16a3234
08 Aug, 2018 3 commits
- add support for early stopping (#59) · 7606fc3b
  Gao, Xiang authored Aug 08, 2018
  
  7606fc3b
- Rename to better describe what it is (#58) · f93a9ba4
  Gao, Xiang authored Aug 08, 2018
  
  f93a9ba4
- fix sort_by_species (#57) · 5bb66915
  Gao, Xiang authored Aug 07, 2018
  
  5bb66915
07 Aug, 2018 2 commits
- a new example to test neurochem on specified dataset (#56) · e5439f3d
  Gao, Xiang authored Aug 07, 2018
  
  e5439f3d
- log training rmse and best validation rmse · 7f660099
  Gao, Xiang authored Aug 07, 2018
  
  7f660099
03 Aug, 2018 5 commits
- checkpoint if validation gets better result (#55) · 3e296069
  Gao, Xiang authored Aug 03, 2018
  
  3e296069
- remove output_length (#54) · f50cc0b4
  Gao, Xiang authored Aug 03, 2018
  
  f50cc0b4
- improve energy shifter (#52) · 73e447f0
  Gao, Xiang authored Aug 03, 2018
  
  73e447f0
- implement per atom loss (#51) · 59b31d84
  Gao, Xiang authored Aug 03, 2018
  
  59b31d84
- merge Container and BatchModel to provide richer information in output (#50) · 84439caf
  Gao, Xiang authored Aug 03, 2018
  
  84439caf
02 Aug, 2018 3 commits
- add argparser for nnp_training.py (#46) · 1d8bba37
  Gao, Xiang authored Aug 02, 2018
  
  1d8bba37
- add argparse to training-benchmark.py (#45) · 861580e3
  Gao, Xiang authored Aug 02, 2018
  
  861580e3
- remove explicit device and dtype (#44) · d3ae0788
  Gao, Xiang authored Aug 02, 2018
  
  d3ae0788
01 Aug, 2018 2 commits
- use 0 size n dim tensor feature to simplify code (#43) · 8c493a6e
  Gao, Xiang authored Aug 01, 2018
  
  8c493a6e
- use pytorch's CELU to simplify code (#42) · 0e992fe5
  Gao, Xiang authored Aug 01, 2018
  
  0e992fe5
31 Jul, 2018 9 commits
- fix build (#41) · 97e3df07
  Gao, Xiang authored Jul 31, 2018
  
  97e3df07
- completely decouple aev computer and neural networks (#40) · e2e740d0
  Gao, Xiang authored Jul 31, 2018
  
  e2e740d0
- remove force from ANIModel (#39) · ee31b752
  Gao, Xiang authored Jul 31, 2018
  
  ee31b752
- make aev computer accept species as tensor (#38) · 2cb357bd
  Gao, Xiang authored Jul 31, 2018
  
  2cb357bd
- make aev computer take a single tuple as input (#37) · 629bc698
  Gao, Xiang authored Jul 31, 2018
  
  629bc698
- add tqdm and tensorboard support on example (#35) · d07bd02c
  Gao, Xiang authored Jul 31, 2018
  
  d07bd02c
- test energy force example (#34) · f6b193bd
  Gao, Xiang authored Jul 30, 2018
  
  f6b193bd
- assert over-fitting in test_ignite (#33) · 45f25ec3
  Gao, Xiang authored Jul 30, 2018
  
  45f25ec3
- working ignite and dataloader with simple examples (#21) · b3744935
  Gao, Xiang authored Jul 30, 2018
  
  b3744935
30 Jul, 2018 3 commits
- add common loss and metrics (#32) · ea718be0
  Gao, Xiang authored Jul 30, 2018
  
  ea718be0
- add DictMetric (#31) · e2bc9f29
  Gao, Xiang authored Jul 30, 2018
  
  e2bc9f29
- allow energy shifter as transformations to dataset (#30) · 10699bf7
  Gao, Xiang authored Jul 30, 2018
  
  10699bf7
29 Jul, 2018 1 commit
- Add ignite API helpers, also temporarily disable JIT and python2 compatibiltiy (#29) · 18e4867d
  Gao, Xiang authored Jul 29, 2018
  
  18e4867d
28 Jul, 2018 1 commit
- helper function to create dataloader, helper module that handle batch (#27) · 965efee2
  Gao, Xiang authored Jul 28, 2018
  
  965efee2
24 Jul, 2018 2 commits
- Delete nn.py and create torchani.models to support rich types of models (#26) · 194e88ff
  Gao, Xiang authored Jul 24, 2018
  
  194e88ff
- rewrite dataloader API to improve performance and reduce code complexity (#18) · 327a9b20
  Gao, Xiang authored Jul 24, 2018
  
  327a9b20
23 Jul, 2018 1 commit
- add unit test for model ensemble (#14) · f6ef4ebb
  Gao, Xiang authored Jul 23, 2018
  
  f6ef4ebb
22 Jul, 2018 2 commits
- add limited python2 compatibility (#13) · ba3036d1
  Gao, Xiang authored Jul 22, 2018
  
  ba3036d1
- add style check to unit test (#12) · 87c4184a
  Gao, Xiang authored Jul 22, 2018
  
  87c4184a