Commits · 18b26ec66ce829ec8f51ba1322fd8837331dd41c · OpenDAS / Megatron-LM

03 Apr, 2023 5 commits
- Save word embeddings for head only only if embeddings are not untied · 18b26ec6
  Sandeep Subramanian authored Feb 15, 2023
  
  18b26ec6
- disable embedding addreduce if untie_embeddings_and_output_weights · f11b4c99
  Jimmy Zhang authored Jan 31, 2023
  
  f11b4c99
- Add assert for pp and untie · a3fbac58
  MaximumEntropy authored Dec 08, 2022
```
Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca>
```
  a3fbac58
- Fix checkpointing · 36c3a1c1
  MaximumEntropy authored Dec 08, 2022
```
Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca>
```
  36c3a1c1
- Initial commit for untied embeddings · 82c7ba57
  MaximumEntropy authored Dec 07, 2022
```
Signed-off-by: MaximumEntropy <sandeep.subramanian.1@umontreal.ca>
```
  82c7ba57
01 Apr, 2023 2 commits
- Merge branch 'core-pipeline-fix' into 'main' · fdd34a82
  Jared Casper authored Apr 01, 2023
```
Fix bug in core pipeline schedule

See merge request ADLR/megatron-lm!552
```
  fdd34a82
- Fix bug in core pipeline schedule · ff7b28f4
  Jared Casper authored Apr 01, 2023
  
  ff7b28f4
31 Mar, 2023 2 commits
- Merge branch 'activation-checkpoint-fix' into 'main' · 2c493fb3
  Jared Casper authored Mar 31, 2023
```
fix bug in uniform activation recompute

See merge request ADLR/megatron-lm!551
```
  2c493fb3
- fix bug in uniform activation recompute · fc2c81d3
  Jared Casper authored Mar 31, 2023
  
  fc2c81d3
27 Mar, 2023 2 commits
- Merge branch 'maanug/print-logdir' into 'main' · 035cae2e
  Maanu Grover authored Mar 27, 2023
```
Print result logs directory if Pytest fails

See merge request ADLR/megatron-lm!549
```
  035cae2e
- Print result logs directory if Pytest fails · d86299b0
  Maanu Grover authored Mar 27, 2023
  
  d86299b0
23 Mar, 2023 6 commits
- Merge branch 'core-pipeline' into 'main' · edca71da
  Jared Casper authored Mar 23, 2023
```
Move pipeline parallel functionality into core with associated changes.

See merge request ADLR/megatron-lm!499
```
  edca71da
- Move pipeline parallel functionality into core with associated changes. · 3c92fa93
  Jared Casper authored Mar 23, 2023
  
  3c92fa93
- Merge branch 'maanug/unit-tests-rules' into 'main' · 0b44909c
  Jared Casper authored Mar 22, 2023
```
Add rules for unit tests

See merge request ADLR/megatron-lm!545
```
  0b44909c
- Add rules for unit tests · d7d7591d
  Maanu Grover authored Mar 22, 2023
  
  d7d7591d
- Merge branch 'mausin/contiguous' into 'main' · a0b8ab23
  Jared Casper authored Mar 22, 2023
```
Add contiguous to grad_output in layers.py

See merge request ADLR/megatron-lm!542
```
  a0b8ab23
- Add contiguous to grad_output in layers.py · 09c2f6e4
  Markel Ausin authored Mar 22, 2023
  
  09c2f6e4
22 Mar, 2023 2 commits
- Merge branch 'BugFix' into 'main' · 5b78040e
  Jared Casper authored Mar 22, 2023
```
Update .gitlab-ci.yml

See merge request ADLR/megatron-lm!544
```
  5b78040e
- Update .gitlab-ci.yml · 7d1d560c
  Shanmugam Ramasamy authored Mar 22, 2023
  
  7d1d560c
16 Mar, 2023 4 commits
- Merge branch 'merge_req' into 'main' · b7f3c263
  Jared Casper authored Mar 16, 2023
```
Update .gitlab-ci.yml

See merge request ADLR/megatron-lm!540
```
  b7f3c263
- Update .gitlab-ci.yml · 6289787b
  Shanmugam Ramasamy authored Mar 16, 2023
  
  6289787b
- Merge branch 'UpdateTests' into 'main' · 7aceba87
  Shanmugam Ramasamy authored Mar 16, 2023
```
Update .gitlab-ci.yml

See merge request ADLR/megatron-lm!537
```
  7aceba87
- Update .gitlab-ci.yml · 9eb04e96
  Shanmugam Ramasamy authored Mar 16, 2023
  
  9eb04e96
14 Mar, 2023 2 commits
- Merge branch 'bugFixRestarting' into 'main' · 21c19e81
  Shanmugam Ramasamy authored Mar 14, 2023
```
Update .gitlab-ci.yml to support restarting pipeline

See merge request ADLR/megatron-lm!534
```
  21c19e81
- Update .gitlab-ci.yml to support restarting pipeline · 89a9f399
  Shanmugam Ramasamy authored Mar 14, 2023
  
  89a9f399
09 Mar, 2023 2 commits
- Merge branch 'maanug/gpt-text-gen' into 'main' · 636da044
  Jared Casper authored Mar 09, 2023
```
Fix GPT text generation

See merge request ADLR/megatron-lm!528
```
  636da044
- Fix GPT text generation · 3c760180
  Maanu Grover authored Mar 09, 2023
  
  3c760180
08 Mar, 2023 2 commits
- Merge branch 'citest' into 'main' · f5cf2e42
  Jared Casper authored Mar 08, 2023
```
Testing infrastructure for Megatron core

See merge request ADLR/megatron-lm!514
```
  f5cf2e42
- Testing infrastructure for Megatron core · 8e6fa622
  Shanmugam Ramasamy authored Mar 08, 2023
  
  8e6fa622
22 Feb, 2023 6 commits
- Merge branch 'maanug/readme-pretraining' into 'main' · 95f872f5
  Jared Casper authored Feb 21, 2023
```
Pretraining README update

See merge request ADLR/megatron-lm!512
```
  95f872f5
- Pretraining README update · 5ecd52a4
  Maanu Grover authored Feb 21, 2023
  
  5ecd52a4
- Merge branch 'maanug/paper-links' into 'main' · f063bbb5
  Jared Casper authored Feb 21, 2023
```
README: Paper Links

See merge request ADLR/megatron-lm!521
```
  f063bbb5
- README: Paper Links · 23a54e02
  Maanu Grover authored Feb 21, 2023
  
  23a54e02
- Merge branch 'lmcafee/retro' into 'main' · 762072e0
  Jared Casper authored Feb 21, 2023
```
Retro

See merge request ADLR/megatron-lm!489
```
  762072e0
- Retro · 17a6044d
  Lawrence McAfee authored Feb 21, 2023
  
  17a6044d
17 Feb, 2023 5 commits
- Merge branch 'maanug/distributed-pretrain-examples' into 'main' · ef59b687
  Jared Casper authored Feb 17, 2023
```
Fix distributed pretraining examples

See merge request ADLR/megatron-lm!523
```
  ef59b687
- Fix distributed pretraining examples · 94dbfd1c
  Maanu Grover authored Feb 17, 2023
  
  94dbfd1c
- Merge branch 'maanug/readme-pretraining-dist' into 'main' · 0dce440a
  Jared Casper authored Feb 17, 2023
```
Update Distributed Pretraining REAME section

See merge request ADLR/megatron-lm!522
```
  0dce440a
- Update Distributed Pretraining REAME section · 912ab8ad
  Maanu Grover authored Feb 17, 2023
  
  912ab8ad
- Merge branch 'maanug/1gpu-pretrain-examples' into 'main' · 0b208deb
  Jared Casper authored Feb 17, 2023
```
Fix pretraining examples

See merge request ADLR/megatron-lm!513
```
  0b208deb