Commits · ceca097fc54bce8deeffbb29ccbde98396ab5310 · OpenDAS / apex

15 Jun, 2020 1 commit
- editing comments · ceca097f
  Asit authored Jun 15, 2020
  
  ceca097f
11 Jun, 2020 2 commits
- Merge pull request #883 from NVIDIA/schetlur/stream_bug_fix · 02a33875
  schetlur authored Jun 11, 2020
```
Update softmax.h
```
  02a33875
- Update softmax.h · 060bd5cd
  schetlur authored Jun 11, 2020
```
Fixing NULL stream launch in dispatch_additive_masked_softmax
```
  060bd5cd
10 Jun, 2020 8 commits
- Merge pull request #880 from seryilmaz/seryilmaz/stream · 058addbe
  Cliff Woolley authored Jun 10, 2020
```
add streaming support for softmax kernels
```
  058addbe
- add streamid · 1574c03d
  Sukru Eryilmaz authored Jun 10, 2020
  
  1574c03d
- add streaming suppport for softmax kernels · 16869317
  Sukru Eryilmaz authored Jun 10, 2020
  
  16869317
- Merge pull request #878 from ksivaman/sparsity · 3e474e85
  Thor Johnsen authored Jun 10, 2020
```
Sparsity
```
  3e474e85
- fixed import error · 73666989
  Kirthi Sivamani authored Jun 10, 2020
  
  73666989
- update README · b152a8dd
  Kirthi Sivamani authored Jun 10, 2020
  
  b152a8dd
- added README for ASP API · 3028b18b
  Kirthi Sivamani authored Jun 10, 2020
  
  3028b18b
- asp files · 2a5b726a
  Kirthi Sivamani authored Jun 10, 2020
  
  2a5b726a
01 Jun, 2020 3 commits
- Add Pyprof removal warnings that point to new repo (#862) · 097238f8
  mcarilli authored Jun 01, 2020
```
Co-authored-by: Michael Carilli <mcarilli@nvidia.com>
```
  097238f8
- Merge pull request #867 from NVIDIA/fix_python_only_apex_distopt · 76026a35
  Thor Johnsen authored Jun 01, 2020
```
Remove distributed lamb from __init__.py
```
  76026a35
- Remove distributed lamb from __init__.py · 012135b1
  Thor Johnsen authored Jun 01, 2020
  
  012135b1
31 May, 2020 6 commits
- Merge pull request #863 from NVIDIA/distributed_lamb_optimizer · cdb0b3ee
  Thor Johnsen authored May 30, 2020
```
Distributed lamb optimizer
```
  cdb0b3ee
- Remove temporary file · 55ae5fbb
  Thor Johnsen authored May 30, 2020
  
  55ae5fbb
- Remove verbose print-out · b614a855
  Thor Johnsen authored May 30, 2020
  
  b614a855
- WIP · ba453a8e
  Thor Johnsen authored May 31, 2020
  
  ba453a8e
- Delayed init · 75f1e9d7
  Thor Johnsen authored May 31, 2020
  
  75f1e9d7
- Delayed init · 53cfd8c2
  Thor Johnsen authored May 30, 2020
  
  53cfd8c2
30 May, 2020 19 commits
- Make step global state variable · 4c54fd2b
  Thor Johnsen authored May 30, 2020
  
  4c54fd2b
- Make step global state variable · 5caf95ca
  Thor Johnsen authored May 30, 2020
  
  5caf95ca
- Bug fix · 7741808b
  Thor Johnsen authored May 30, 2020
  
  7741808b
- Add more default values from regular lamb optimizer · 12458152
  Thor Johnsen authored May 30, 2020
  
  12458152
- Bug fix in update norm calculation · 1e0aadd5
  Thor Johnsen authored May 30, 2020
  
  1e0aadd5
- Bug fix in update norm calculation · 0f64f6ad
  Thor Johnsen authored May 30, 2020
  
  0f64f6ad
- Bug fix in update norm calculation · 56650eb8
  Thor Johnsen authored May 30, 2020
  
  56650eb8
- Bug fix · fb2d0f48
  Thor Johnsen authored May 30, 2020
  
  fb2d0f48
- Bug fix · 3c02784b
  Thor Johnsen authored May 30, 2020
  
  3c02784b
- Bug fix · 9773218c
  Thor Johnsen authored May 30, 2020
  
  9773218c
- Add optional accumulation step · 02fd7341
  Thor Johnsen authored May 30, 2020
  
  02fd7341
- Bug fix · 9a09107c
  Thor Johnsen authored May 30, 2020
  
  9a09107c
- Bug fix · 1ac28972
  Thor Johnsen authored May 30, 2020
  
  1ac28972
- Bug fix · ef437358
  Thor Johnsen authored May 30, 2020
  
  ef437358
- Bug fix · e6925e6c
  Thor Johnsen authored May 30, 2020
  
  e6925e6c
- Use correct names for mt lamb cuda kernels · 8ed8eaac
  Thor Johnsen authored May 30, 2020
  
  8ed8eaac
- Make separate apex option for distributed lamb · 45388d48
  Thor Johnsen authored May 30, 2020
  
  45388d48
- Remove unused forward def · 848a2844
  Thor Johnsen authored May 30, 2020
  
  848a2844
- Distributed LAMB optimizer · 19892f1d
  Thor Johnsen authored May 30, 2020
  
  19892f1d
29 May, 2020 1 commit
- Fixes to Multihead Attention with LayerNorm and Dropout-Add (#860) · 5754fa7a
  Kevin Stephano authored May 29, 2020
  
  5754fa7a