Commits · 52a5f2f272e5ef242eb271227da712f7dfc55da3 · OpenDAS / Megatron-LM

12 Nov, 2020 3 commits
- Intra-layer MP -> Tensor MP, Inter-layer MP -> Pipeline MP · 52a5f2f2
  Deepak Narayanan authored Oct 20, 2020
  
  52a5f2f2
- Pipeline parallelism implementation with periodic full-pipeline syncs · 7abd3e90
  Deepak Narayanan authored Aug 29, 2020
```
Also includes following changes for inter-layer model-parallel implementation:
- Refactoring of model implementations
- Training loop changes to support inter-layer communication using `ring_exchange`
- New groups for inter-layer communication
- Checkpoint changes
- Command line arguments
```
  7abd3e90
- fp32 working · 28cd66e1
  mohammad authored Aug 28, 2020
  
  28cd66e1
03 Nov, 2020 2 commits
- Merge branch 'fix_logging' into 'main' · b4b0d739
  Deepak Narayanan authored Nov 03, 2020
```
fixed loss average when all but one value is skipped

See merge request ADLR/megatron-lm!164
```
  b4b0d739
- fixed loss average when all but one value is skipped · 664cd28b
  mohammad authored Nov 03, 2020
  
  664cd28b
14 Oct, 2020 6 commits
- Merge branch 'main_evaluate_wiki' into 'main' · 79888e16
  Mohammad Shoeybi authored Oct 14, 2020
```
fixed wiki evaluation

See merge request ADLR/megatron-lm!157
```
  79888e16
- fixed wiki evaluation issue · ef2adb5d
  Mostofa Patwary authored Oct 14, 2020
  
  ef2adb5d
- fixed wiki evaluation issue · 38c45de7
  Mostofa Patwary authored Oct 14, 2020
  
  38c45de7
- Merge branch 'main_beta' into 'main' · 64cf3d98
  Mohammad Shoeybi authored Oct 13, 2020
```
Adam betas and eps

See merge request ADLR/megatron-lm!156
```
  64cf3d98
- added adam betas and eps as arguments · c55e154f
  Mostofa Patwary authored Oct 13, 2020
  
  c55e154f
- added adam betas and eps as arguments · 48269d8d
  Mostofa Patwary authored Oct 13, 2020
  
  48269d8d
13 Oct, 2020 5 commits
- Merge branch 'vijay/suppress_ninja_output' into 'main' · 5753e8f1
  Mohammad Shoeybi authored Oct 12, 2020
```
suppress kernel loading output

See merge request ADLR/megatron-lm!155
```
  5753e8f1
- suppress kernel loading output · 8fca9b49
  Vijay Korthikanti authored Oct 12, 2020
  
  8fca9b49
- Merge branch 'vijay/ordering_perf_fix' into 'main' · 538f0e05
  Mohammad Shoeybi authored Oct 12, 2020
```
reordering perf fix

See merge request ADLR/megatron-lm!154
```
  538f0e05
- typo fix · 01dffdb5
  Vijay Korthikanti authored Oct 12, 2020
  
  01dffdb5
- address review comments · 0bc75448
  Vijay Korthikanti authored Oct 12, 2020
  
  0bc75448
12 Oct, 2020 2 commits
- Incrementing checkpoint version to 2.0 · 9b0083ea
  Vijay Korthikanti authored Oct 12, 2020
  
  9b0083ea
- reordering perf fix · ee327acd
  Vijay Korthikanti authored Oct 12, 2020
  
  ee327acd
09 Oct, 2020 3 commits
- Merge branch 'build_fix' into 'main' · 42d21122
  Mohammad Shoeybi authored Oct 09, 2020
```
Create build directory for fused_kernels before building.

See merge request ADLR/megatron-lm!149
```
  42d21122
- Merge branch 'vijay/diff_query_key_lengths' into 'main' · 54282071
  Mohammad Shoeybi authored Oct 08, 2020
```
support for different query key sequence lengths

See merge request ADLR/megatron-lm!151
```
  54282071
- support for different query key sequence lengths · d763f54a
  Vijay Korthikanti authored Oct 08, 2020
  
  d763f54a
08 Oct, 2020 1 commit
- Create build directory for fused_kernels before building. · 54ded172
  Jared Casper authored Oct 07, 2020
  
  54ded172
06 Oct, 2020 2 commits
- Merge branch 'vijay/fused_kernel_compiliation' into 'main' · 56e16cba
  Jared Casper authored Oct 05, 2020
```
Avoid non deterministic arch order in compilation

See merge request ADLR/megatron-lm!146
```
  56e16cba
- build directory · 3dcd7dc0
  Vijay Korthikanti authored Oct 05, 2020
  
  3dcd7dc0
05 Oct, 2020 2 commits
- build_directory argument · a7f4e460
  Vijay Korthikanti authored Oct 05, 2020
  
  a7f4e460
- avoid non deterministic arch order in compilation · 37d77ca0
  Vijay Korthikanti authored Oct 05, 2020
  
  37d77ca0
02 Oct, 2020 5 commits
- Merge branch 'pip_1.1.5' into 'main' · d80433e1
  Mohammad Shoeybi authored Oct 02, 2020
```
pip 1.1.5

See merge request ADLR/megatron-lm!144
```
  d80433e1
- pip 1.1.5 · f9387ec3
  Evelina authored Oct 02, 2020
```
Signed-off-by: Evelina <ebakhturina@nvidia.com>
```
  f9387ec3
- Merge branch 'pip_1.1.4' into 'main' · b7dac89a
  Mohammad Shoeybi authored Oct 01, 2020
```
1.1.4. pip version

See merge request ADLR/megatron-lm!120
```
  b7dac89a
- Merge branch 'mpatwary/main_text_gen' into 'main' · b1870045
  Mohammad Shoeybi authored Oct 01, 2020
```
Fixed uninit communication gruop

See merge request ADLR/megatron-lm!142
```
  b1870045
- Fixed uninit communication gruop · 057a1fab
  Mostofa Patwary authored Oct 01, 2020
  
  057a1fab
01 Oct, 2020 2 commits
- Merge branch 'check_mpu_init' into 'main' · 10c828ed
  Jared Casper authored Oct 01, 2020
```
fix for nemo: do not initialize mpu if it is already initialized

See merge request ADLR/megatron-lm!141
```
  10c828ed
- fix for nemo: do not initialize mpu if it is already initialized · 6e433055
  mohammad authored Oct 01, 2020
  
  6e433055
30 Sep, 2020 3 commits
- Merge branch 'vijay/weight_mapping' into 'main' · c63906a6
  Mohammad Shoeybi authored Sep 29, 2020
```
Vijay/weight mapping

See merge request ADLR/megatron-lm!138
```
  c63906a6
- address review comments · 64e45f29
  Vijay Korthikanti authored Sep 29, 2020
  
  64e45f29
- checkpoint versioning · a8d47812
  Vijay Korthikanti authored Sep 29, 2020
  
  a8d47812
29 Sep, 2020 4 commits
- Merge branch 'main_megatron_cntrl' into 'main' · 577b4657
  Mohammad Shoeybi authored Sep 29, 2020
```
Text generation fixes

See merge request ADLR/megatron-lm!137
```
  577b4657
- bug fix · 7d4ad51e
  Vijay Korthikanti authored Sep 29, 2020
  
  7d4ad51e
- fix of few uninitialized varibles · 826d2452
  Mostofa Patwary authored Sep 29, 2020
  
  826d2452
- support for old checkpoint format · 37ae6646
  Vijay Korthikanti authored Sep 28, 2020
  
  37ae6646