Commits · 229db8c695249a86775099aec383bfb3a061f180 · 杨迪 / NiuTrans.Tensor

23 Nov, 2018 1 commit
- improve the implementation of Unsqueeze · 229db8c6
  xiaotong committed Nov 23, 2018
  
  229db8c6 Browse Files
22 Nov, 2018 1 commit
- fix the bug of invalid memory pointer in releasing a memory pool · fe90b454
  xiaotong committed Nov 22, 2018
  
  fe90b454 Browse Files
21 Nov, 2018 3 commits
- new comment · 48661938
  xiaotong committed Nov 21, 2018
  
  48661938 Browse Files
- improve the implementation of _CudaSpreadForGather · 314a02af
  xiaotong committed Nov 21, 2018
  
  314a02af Browse Files
- improve the code of _CudaBLASMatrixMULList · 8f665e61
  xiaotong committed Nov 21, 2018
  
  8f665e61 Browse Files
19 Nov, 2018 3 commits
- remove some code · ac620226
  xiaotong committed Nov 19, 2018
  
  ac620226 Browse Files
- Merge branch 'xuchen' into xiaotong-working · 80ab70a2
  xuchen committed Nov 19, 2018
  
  80ab70a2 Browse Files
- restore the previous implemention · b83a6798
  xuchen committed Nov 19, 2018
  
  b83a6798 Browse Files
18 Nov, 2018 1 commit
- use smaller batches for training · 411cff4c
  xiaotong committed Nov 18, 2018
  
  411cff4c Browse Files
17 Nov, 2018 1 commit
- remove model dumping · 383759ae
  xiaotong committed Nov 17, 2018
  
  383759ae Browse Files
16 Nov, 2018 2 commits
- move the hard dump of model · af3acf68
  xiaotong committed Nov 16, 2018
  
  af3acf68 Browse Files
- correct the bug of loading samples for Tranformer training · 78339123
  xiaotong committed Nov 16, 2018
  
  78339123 Browse Files
15 Nov, 2018 2 commits
- reduce memory footprint for XMem · ca343382
  xiaotong committed Nov 15, 2018
  
  ca343382 Browse Files
- debugging · 12f84346
  xiaotong committed Nov 15, 2018
  
  12f84346 Browse Files
14 Nov, 2018 4 commits
- dump mem info · fb2bdabe
  xiaotong committed Nov 14, 2018
  
  fb2bdabe Browse Files
- dump info · 1037cc81
  xiaotong committed Nov 14, 2018
  
  1037cc81 Browse Files
- gather implementation in mt · c44d1a79
  xuchen committed Nov 14, 2018
  
  c44d1a79 Browse Files
- gather implemention in mt · a400b619
  xuchen committed Nov 14, 2018
  
  a400b619 Browse Files
13 Nov, 2018 2 commits
- Merge branch 'xuchen' into xiaotong-working · 7809ed05
  xuchen committed Nov 13, 2018
  
  7809ed05 Browse Files
- 1. add some base functions 2.better implementation for t2t · 03a9836e
  xuchen committed Nov 13, 2018
  
  03a9836e Browse Files
18 Oct, 2018 1 commit
- correct output length for dumping · f4be1882
  xiaotong committed Oct 18, 2018
  
  f4be1882 Browse Files
17 Oct, 2018 1 commit
- bug fixes of the input of the decoder for t2t · 12195a67
  xiaotong committed Oct 17, 2018
  
  12195a67 Browse Files
15 Oct, 2018 2 commits
- add doubledend flat for t2t mt · 645c32dc
  xiaotong committed Oct 15, 2018
  
  645c32dc Browse Files
- add padding on the encoder side for t2t MT · 35e084b0
  xiaotong committed Oct 15, 2018
  
  35e084b0 Browse Files
11 Oct, 2018 1 commit
- fix the bug in sorting sequences · 6f90577d
  xiaotong committed Oct 11, 2018
  
  6f90577d Browse Files
10 Oct, 2018 6 commits
- bug fixes for t2t mt · a037d802
  xiaotong committed Oct 10, 2018
  
  a037d802 Browse Files
- Merge branch 'xiaotong-working' of 47.105.50.196:NiuTrans/NiuTrans.Tensor into xiaotong-working · ecc4041d
  xiaotong committed Oct 10, 2018
  
  ecc4041d Browse Files
- local machine version · 4a609624
  xiaotong committed Oct 10, 2018
  
  4a609624 Browse Files
- comment · 03064396
  xiaotong committed Oct 10, 2018
  
  03064396 Browse Files
- improve initialization · bd37527b
  xiaotong committed Oct 10, 2018
  
  bd37527b Browse Files
- fix bugs and improve the implementation in the CrossEntropy function · 52c0e35a
  xuchen committed Oct 10, 2018
  
  52c0e35a Browse Files
09 Oct, 2018 2 commits
- load batch of sequence on both langauge sides · c6f50a22
  xiaotong committed Oct 09, 2018
  
  c6f50a22 Browse Files
- add the decoder for transformer · 430f0dfc
  xiaotong committed Oct 09, 2018
  
  430f0dfc Browse Files
02 Oct, 2018 3 commits

fix a bug in the softmax function. It executes backward comptuation incorrectly… · b409f07f

fix a bug in the softmax function. It executes backward comptuation incorrectly (in the condition of cpu computation and NOLOSS).
I found this bug by chance, then I spent one night confirming that something must be wrong and I spent one night finding the location of bug.
Finally, I fixed this bug in five minutes.

committed Oct 03, 2018

b409f07f Browse Files

correct the name · 7250ec45
xiaotong committed Oct 02, 2018

7250ec45 Browse Files
clean the code · 6a4741bc
xiaotong committed Oct 02, 2018

6a4741bc Browse Files

01 Oct, 2018 3 commits
- Merge branch 'xuchen' into xiaotong-working · d992a0e9
  xuchen committed Oct 02, 2018
  
  d992a0e9 Browse Files
- 1. add gather function 2. add cross entropy forward computation and backward… · ceb5b101
```
1. add gather function 2. add cross entropy forward computation and backward computation 3. code optimization 4. merge with xiaotong-working branch
```
  xuchen committed Oct 02, 2018
  ceb5b101 Browse Files
- remove source.zip · 7df7bc1d
  xiaotong committed Oct 01, 2018
  
  7df7bc1d Browse Files
30 Sep, 2018 1 commit
- normalize the loss · 1e3c89a0
  xiaotong committed Sep 30, 2018
  
  1e3c89a0 Browse Files