Commits · master · Anthony J. Wimmers / DeepLearning

Aug 29, 2019
- Trying again to make work with gpu · a13d0494
  Tony Wimmers authored 5 years ago
  
  a13d0494
Aug 28, 2019
- Full, not unit · 98f0a32d
  Tony Wimmers authored 5 years ago
  
  98f0a32d
- unit testing new directories · ad5538d5
  Tony Wimmers authored 5 years ago
  
  ad5538d5
- loading modules correctly, I think · 3e916f27
  Tony Wimmers authored 5 years ago
  
  3e916f27
- loading modules correctly, I think · 0ff983aa
  Tony Wimmers authored 5 years ago
  
  0ff983aa
- loading modules correctly, I think · 8d3815bc
  Tony Wimmers authored 5 years ago
  
  8d3815bc
Aug 24, 2019
- simplify devices more · fba2e0a9
  Tony Wimmers authored 5 years ago
  
  fba2e0a9
Aug 23, 2019
- Simplify devices for slurm · aa223af0
  Tony Wimmers authored 5 years ago
  
  aa223af0
- changing paths · 3abcf404
  Tony Wimmers authored 5 years ago
  
  3abcf404
- changing paths · 52461245
  Tony Wimmers authored 5 years ago
  
  52461245
- Tonys initial commit · 76c52f7c
  Tony Wimmers authored 5 years ago
  
  76c52f7c
Aug 20, 2019
- minor · 329df464
  rink authored 5 years ago
  
  329df464
Aug 16, 2019
- minor · ef1406d4
  rink authored 5 years ago
  
  ef1406d4
Aug 13, 2019
- make restore methods better · 55c2bead
  rink authored 5 years ago
  
  55c2bead
- minor · 4118f8e0
  rink authored 5 years ago
  
  4118f8e0
Aug 12, 2019
- Better handling of saved models · 559540c5
  rink authored 5 years ago
  
  559540c5
Aug 08, 2019
- match batch_size, learning_rate and max num epochs to latest good params · 863775ff
  rink authored 5 years ago
  
  863775ff
Aug 06, 2019
- minor · 1b3810fc
  rink authored 5 years ago
  
  1b3810fc
- bump up learning rate to 0.0180 · f29404c4
  rink authored 5 years ago
  
  f29404c4
- minor · 73add11c
  rink authored 5 years ago
  
  73add11c
- minor · 5164388d
  rink authored 5 years ago
  
  5164388d
Aug 05, 2019
- minor · 9710ddeb
  rink authored 5 years ago
  
  9710ddeb
- make sure we don't modify the compute graph during training. This can cause · 5cbba284
  rink authored 5 years ago
```
meta_graph files grow throughout the process.
```
  5cbba284
Aug 02, 2019
- minor · d8687d1d
  rink authored 5 years ago
  
  d8687d1d
Aug 01, 2019
- minor · f16eeadf
  rink authored 5 years ago
  
  f16eeadf
- generalize a bit · 7431ba51
  rink authored 5 years ago
  
  7431ba51
Jul 29, 2019
- revert to old model save path · bae8a9d8
  rink authored 5 years ago
  
  bae8a9d8
- (1) Turn off dropout inside the residual blocks. · 027260a7
  rink authored 5 years ago
```
(2) Slightly increase the learning_rate
(3) Run test data through in batches to reduce memory bloat
```
  027260a7
Jul 24, 2019
- have explicit per run directoris for both model and log saving · b037be2e
  rink authored 5 years ago
  
  b037be2e
Jul 22, 2019
- minor · b8d51700
  rink authored 5 years ago
  
  b8d51700
- minor · b2957f85
  rink authored 5 years ago
  
  b2957f85
- make sure we do all evaluation with test data in batches so we don't run · 2ff2fc97
  rink authored 5 years ago
```
out of memory.
```
  2ff2fc97
Jul 20, 2019
- minor · a653aff4
  rink authored 5 years ago
  
  a653aff4
- minor · a5ad9a9c
  Tom Rink authored 5 years ago
  
  a5ad9a9c
- make all layers 60 wide · 0d63d1f0
  Tom Rink authored 5 years ago
  
  0d63d1f0
Jul 19, 2019
- fix typo · 0c9a4035
  Tom Rink authored 5 years ago
  
  0c9a4035
- add another layer with a residual block · 302a7c99
  Tom Rink authored 5 years ago
  
  302a7c99
- add a second residual block · 58da87ae
  Tom Rink authored 5 years ago
  
  58da87ae
- minor · 67ef55ee
  Tom Rink authored 5 years ago
  
  67ef55ee
- add some stats to restore method · 2f3a1e8c
  Tom Rink authored 5 years ago
  
  2f3a1e8c