Commits · 3c2a59790566cd03ffd869301b49f9997bec9937 · OpenDAS / FastMoE

"examples/community/wildcard_stable_diffusion.py" did not exist on "bd8df2da89d99f630e5aa2ddb8f8cb45456561f1"

22 Mar, 2021 11 commits
- Merge pull request #19 from laekov/balance · 3c2a5979
  Rick Ho authored Mar 22, 2021
```
Add balance strategy
```
  3c2a5979
- Revise variable name · e028f2ec
  Sengxian authored Mar 22, 2021
  
  e028f2ec
- Merge remote-tracking branch 'origin/master' into balance_public · 69121432
  Sengxian authored Mar 22, 2021
  
  69121432
- Merge pull request #18 from laekov/checkpoint · c1e67585
  Rick Ho authored Mar 22, 2021
```
Checkpoint
```
  c1e67585
- Merge remote-tracking branch 'origin/master' into checkpoint · d155925d
  Jiezhong Qiu authored Mar 22, 2021
```
Conflicts:
	fmoe/transformer.py
```
  d155925d
- Add gshard and noisy gate balance strategy · 89d6c794
  Sengxian authored Mar 22, 2021
  
  89d6c794
- Update issue templates · 98b4b0be
  Rick Ho authored Mar 22, 2021
  
  98b4b0be
- remove debug info · f49a9ec4
  Jiezhong Qiu authored Mar 22, 2021
  
  f49a9ec4
- format with black · d30b746a
  Jiezhong Qiu authored Mar 22, 2021
  
  d30b746a
- relatex max line length to 120 · 9d718939
  Jiezhong Qiu authored Mar 22, 2021
  
  9d718939
- fix pylint · a372eb66
  Jiezhong Qiu authored Mar 22, 2021
  
  a372eb66
21 Mar, 2021 1 commit
- load/save fp32 main weight when fp16 training · d8124b80
  Jiezhong Qiu authored Mar 22, 2021
  
  d8124b80
17 Mar, 2021 5 commits
- fp16/fp32 optimizer has different data structure · 59f10bf3
  Jiezhong Qiu authored Mar 17, 2021
  
  59f10bf3
- remove unnecessary states when dp rank>0 · f224b69b
  Jiezhong Qiu authored Mar 17, 2021
  
  f224b69b
- Create Megatron.LICENSE · 87c1e6bb
  Jiezhong Qiu authored Mar 17, 2021
  
  87c1e6bb
- load checkpoint for expert parallel · 6ca840b5
  Jiezhong Qiu authored Mar 17, 2021
  
  6ca840b5
- avoid saveing duplicated optim state in dp rank>0 · 1c69da9c
  Jiezhong Qiu authored Mar 17, 2021
  
  1c69da9c
16 Mar, 2021 2 commits
- fix pylint · a54546a4
  Jiezhong Qiu authored Mar 16, 2021
  
  a54546a4
- save experts separately in each data parallel rank · 9c902b82
  Jiezhong Qiu authored Mar 16, 2021
  
  9c902b82
13 Mar, 2021 3 commits
- checkout to v0.1.2 · a3b2eb62
  Rick Ho authored Mar 13, 2021
  
  a3b2eb62
- fix lint · 3b82e379
  Rick Ho authored Mar 13, 2021
  
  3b82e379
- manually broadcast nccl unique id · 43af1522
  Rick Ho authored Mar 13, 2021
  
  43af1522
09 Mar, 2021 1 commit
- Format using black · 4d48209d
  Sengxian authored Mar 09, 2021
  
  4d48209d
08 Mar, 2021 1 commit
- fix numerical issue in tests and typo · 527a8cc9
  Rick Ho authored Mar 08, 2021
  
  527a8cc9
06 Mar, 2021 1 commit
- add ninja in requirements · 12a7921b
  Jiezhong Qiu authored Mar 06, 2021
  
  12a7921b
05 Mar, 2021 1 commit
- fix example relative path in chinese readme · 18798ff9
  Rick Ho authored Mar 05, 2021
  
  18798ff9
04 Mar, 2021 1 commit
- remove dependency on cuda sample repo · 7d695acb
  Rick Ho authored Mar 04, 2021
  
  7d695acb
02 Mar, 2021 1 commit
- fetch the apache license of the transformer-xl example · 7f784c84
  Rick Ho authored Mar 02, 2021
  
  7f784c84
01 Mar, 2021 4 commits
- fix type · 2bcbdb59
  Jiezhong Qiu authored Mar 01, 2021
  
  2bcbdb59
- update a few words in readme · a334037d
  Rick Ho authored Mar 01, 2021
  
  a334037d
- checkout to version v0.1.1 · 0c3aa2cd
  Rick Ho authored Mar 01, 2021
  
  0c3aa2cd
- update megatron example · 53b5b8c3
  Rick Ho authored Mar 01, 2021
  
  53b5b8c3
28 Feb, 2021 8 commits
- bigger logo in title · 2d067240
  Rick Ho authored Feb 28, 2021
  
  2d067240
- update logo in readme · fcd7ee54
  Rick Ho authored Feb 28, 2021
  
  fcd7ee54
- add logo · 4c9bc502
  Rick Ho authored Feb 28, 2021
  
  4c9bc502
- fix pytorch docker link in readme · 2628cedb
  Rick Ho authored Feb 28, 2021
  
  2628cedb
- Chinese readme and release note · dd7cbfc4
  Rick Ho authored Feb 28, 2021
  
  dd7cbfc4
- add slack link to readme · 23d1aa66
  Rick Ho authored Feb 28, 2021
  
  23d1aa66
- adjust figure sizes and positions · 38d34c9c
  Jiezhong Qiu authored Feb 28, 2021
  
  38d34c9c
- Update README.md · 6d2a7bc3
  Jiezhong Qiu authored Feb 28, 2021
  
  6d2a7bc3