Commits · 7890681a8ebd273fc6b1d163c3056c19d97cb097 · OpenDAS / Megatron-LM

"test/test_transforms_v2_functional.py" did not exist on "dfa81ce4191ea8dc96a5b509ed27c6938a8b853d"

03 Apr, 2020 1 commit
- Lint megatron/data/dataset_utils.py · 840759b8
  Neel Kant authored Apr 03, 2020
  
  840759b8
01 Apr, 2020 1 commit
- addressed comments from raul, neel, and jared · bf3ce751
  Mohammad authored Apr 01, 2020
  
  bf3ce751
30 Mar, 2020 4 commits
- removed unused values in samplers · 259062c2
  Mohammad authored Mar 30, 2020
  
  259062c2
- added samplers from data_utils · 32e1ddb0
  Mohammad authored Mar 30, 2020
  
  32e1ddb0
- removed building tokenizer from bert dataset · dedb2ef7
  Mohammad authored Mar 30, 2020
  
  dedb2ef7
- both bert and gpt are working · 5f8623db
  Mohammad authored Mar 29, 2020
  
  5f8623db
29 Mar, 2020 1 commit
- utils partially refactored and tested · 3f58649b
  Mohammad authored Mar 28, 2020
  
  3f58649b
28 Mar, 2020 2 commits
- tokenizer moved to its own directory · 11220df8
  Mohammad authored Mar 28, 2020
  
  11220df8
- arguments.py refactored · 94e2ca57
  Mohammad authored Mar 27, 2020
  
  94e2ca57
27 Mar, 2020 1 commit
- added global variables · 83aa9219
  Mohammad authored Mar 27, 2020
  
  83aa9219
24 Mar, 2020 1 commit
- BERT downstream tasks: RACE, MNLI, QQP · 90ef2e28
  Mohammad Shoeybi authored Mar 23, 2020
  
  90ef2e28
17 Mar, 2020 1 commit
- moved albert to bert · 09e05c6f
  Mohammad Shoeybi authored Mar 16, 2020
  
  09e05c6f
26 Dec, 2019 1 commit
- changed long seq length to 512 · 43dd21f7
  Mohammad Shoeybi authored Dec 26, 2019
  
  43dd21f7
25 Dec, 2019 1 commit
- added filtering based on sentence length · 2c58c9b0
  Mohammad Shoeybi authored Dec 24, 2019
  
  2c58c9b0
04 Dec, 2019 1 commit
- minor cleaning · 75a86a1d
  Mohammad Shoeybi authored Dec 03, 2019
  
  75a86a1d
24 Nov, 2019 4 commits
- removed split dataset · 8179ebd3
  Mohammad Shoeybi authored Nov 24, 2019
  
  8179ebd3
- minor fix · 5300c69f
  Mohammad Shoeybi authored Nov 24, 2019
  
  5300c69f
- resolved reproducibity issue · e2add0fd
  Mohammad Shoeybi authored Nov 24, 2019
  
  e2add0fd
- zero worker seems to be working · 0601702a
  Mohammad Shoeybi authored Nov 24, 2019
  
  0601702a
21 Nov, 2019 1 commit
- fixed padding issue · f6a6811f
  Mohammad Shoeybi authored Nov 20, 2019
  
  f6a6811f
20 Nov, 2019 2 commits
- added write option for index map · 9ea9d50f
  Mohammad Shoeybi authored Nov 20, 2019
  
  9ea9d50f
- fixed couple of bugs in helpers · 690291dd
  Mohammad Shoeybi authored Nov 20, 2019
  
  690291dd
19 Nov, 2019 1 commit
- Plumbing for new albert dataset, including mods to arguments for data loaders. · 1a1cfcff
  Jared Casper authored Nov 18, 2019
  
  1a1cfcff
13 Nov, 2019 1 commit
- Plugging new dataset stuff together. · 0058b1a5
  Jared Casper authored Nov 13, 2019
```
Some cleanup of types, etc. in the helpers cpp code.
```
  0058b1a5
11 Nov, 2019 1 commit
- c++ code working · f51ceb7c
  Mohammad Shoeybi authored Nov 10, 2019
  
  f51ceb7c
09 Nov, 2019 2 commits
- Skip any empty sentences during preprocessing. · 3f4bc91b
  Jared Casper authored Nov 08, 2019
  
  3f4bc91b
- Add ability to take a contiguous slice to indexed_dataset. · 61697cab
  Jared Casper authored Nov 08, 2019
```
Also removed the fix_lua_indexing stuff from fairseq.
```
  61697cab
08 Nov, 2019 5 commits
- before optimization · 6140718f
  Mohammad Shoeybi authored Nov 08, 2019
  
  6140718f
- built simple test for dataset · c125d247
  Mohammad Shoeybi authored Nov 07, 2019
  
  c125d247
- added dataset · 7120e931
  Mohammad Shoeybi authored Nov 07, 2019
  
  7120e931
- added training sample builder · adec01d0
  Mohammad Shoeybi authored Nov 07, 2019
  
  adec01d0
- Add document index to index file. An empty sentence no longer separate documents. · 87bbe9be
  Jared Casper authored Nov 07, 2019
  
  87bbe9be
07 Nov, 2019 2 commits
- Initial commit of multiprocess preprocess and extracted copy of fairseq's indexed_dataset. · 1237533e
  Jared Casper authored Nov 07, 2019
  
  1237533e
- added bert tokenization · 0ceeb3b4
  Mohammad Shoeybi authored Nov 06, 2019
  
  0ceeb3b4