提交 · ced656faaf6a2718a217d11a137b492c00a1b511 · testgroup / pytensor

21 1月, 2016 7 次提交
- Test result of theano function · ced656fa
  由 carriepl 提交于 1月 20, 2016
  
  ced656fa
- Split local_log_softmax_dnn into two separate op lifters · f8bbacbd
  由 carriepl 提交于 1月 20, 2016
  
  f8bbacbd
- Add test for opt to convert LogSoftmax to DnnLogSoftmax in gpuarray backend · 487cf52c
  由 carriepl 提交于 1月 20, 2016
  
  487cf52c
- Add opt to convert LogSoftmax to DnnLogSoftmax in gpuarray backend · da939705
  由 carriepl 提交于 1月 20, 2016
  
  da939705
- Update doctring in gpuarray/dnn.py · 99012195
  由 carriepl 提交于 1月 20, 2016
  
  99012195
- Add test for opt to convert LogSoftmax to DnnLogSoftmax in cuda backend · c40c1ed3
  由 carriepl 提交于 1月 20, 2016
  
  c40c1ed3
- Add opt to convert LogSoftmax to DnnLogSoftmax in cuda backend · bdcb8d26
  由 carriepl 提交于 1月 20, 2016
  
  bdcb8d26
20 1月, 2016 9 次提交
- Merge pull request #3799 from lamblin/avg_pool_v3 · 9a0ab543
  由 Frédéric Bastien 提交于 1月 20, 2016
```
Disable average_exc_pad pooling for cudnn v3
```
  9a0ab543
- Merge pull request #3847 from jojolalpin/master · 136c6d44
  由 Frédéric Bastien 提交于 1月 20, 2016
```
Update opt.py
```
  136c6d44
- Merge pull request #3893 from abergeron/fix_buildbot · f6406742
  由 Pascal Lamblin 提交于 1月 20, 2016
```
Remove more warnings from the buildbot
```
  f6406742
- No error in v3 when padding is of 0 · b4bd7b90
  由 Pascal Lamblin 提交于 1月 19, 2016
  
  b4bd7b90
- Skip tests of average_exc_pad with cudnn v3 · 74b59451
  由 Pascal Lamblin 提交于 1月 19, 2016
  
  74b59451
- Disable average_exc_pad pooling for cudnn v3 · 7b206fb7
  由 Pascal Lamblin 提交于 12月 18, 2015
```
See #3776
```
  7b206fb7
- Get rid of a bit more warnings. · d93640a8
  由 Arnaud Bergeron 提交于 1月 19, 2016
  
  d93640a8
- Follow renaming. · f7136235
  由 Arnaud Bergeron 提交于 1月 19, 2016
  
  f7136235
- Merge pull request #3679 from shabanian/tensor_signal_pep8 · adf81d30
  由 vdumoulin 提交于 1月 19, 2016
```
Renamed max_pool_2d to pool_2d and  DownsampleFactorMax to Pool
```
  adf81d30
19 1月, 2016 16 次提交
- Merge pull request #3884 from abergeron/fix_mac_blas · 6ce3aa78
  由 Pascal Lamblin 提交于 1月 18, 2016
```
Fix blas detection on mac.
```
  6ce3aa78
- Merge pull request #3861 from nouiz/version · ae444d3d
  由 Pascal Lamblin 提交于 1月 18, 2016
```
Bump version to 0.8.0.dev0
```
  ae444d3d
- Merge pull request #3875 from abergeron/fix_buildbot · 3a0493c2
  由 abergeron 提交于 1月 18, 2016
```
Fix buildbot
```
  3a0493c2
- Small doc update. · 534e040c
  由 Arnaud Bergeron 提交于 1月 18, 2016
  
  534e040c
- Flake8 fix. · 3303c991
  由 Arnaud Bergeron 提交于 1月 18, 2016
  
  3303c991
- Tweak the blas flags logic so that it doesn't require certain keys in the dict. · 0c108883
  由 Arnaud Bergeron 提交于 1月 18, 2016
```
Also fix the flags test code so that it doesn't fail due to the mac bug.
```
  0c108883
- Fix brain typo. · a591990a
  由 Arnaud Bergeron 提交于 1月 18, 2016
  
  a591990a
- Use prepare_node() instead of make_thunk() in Elemwise. · 1893c13a
  由 Arnaud Bergeron 提交于 1月 18, 2016
  
  1893c13a
- Add a new prepare_node() method that can be used to do stuff 'just · 710e6b48
  由 Arnaud Bergeron 提交于 1月 18, 2016
```
before' make_thunk and help DebugMode work correctly.
```
  710e6b48
- Add test for the elemwise issue. · 3a0de13a
  由 Arnaud Bergeron 提交于 1月 15, 2016
  
  3a0de13a
- Make cumul be a double variable for better accumulation precision. · e98220f9
  由 Arnaud Bergeron 提交于 1月 15, 2016
  
  e98220f9
- Get rid of some of the deprecation warnings in the buildbot. · ab40a501
  由 Arnaud Bergeron 提交于 1月 14, 2016
  
  ab40a501
- Merge pull request #3363 from fvisin/logsoftmax · 3dcba54d
  由 abergeron 提交于 1月 18, 2016
```
LogSoftmax
```
  3dcba54d
- Merge pull request #2 from abergeron/logsoftmax · 9ad1ea03
  由 Francesco 提交于 1月 18, 2016
```
More complete fix for the travis failures
```
  9ad1ea03
- Remove the try .. except construct since it masked some errors in the opt. · 0041ce67
  由 Arnaud Bergeron 提交于 1月 18, 2016
  
  0041ce67
- Register the logsoftmax opt in fast_compile. · 994c8d35
  由 Arnaud Bergeron 提交于 1月 18, 2016
  
  994c8d35
18 1月, 2016 1 次提交
- Merge pull request #3877 from abergeron/fix_segfault · a5735a1b
  由 Frédéric Bastien 提交于 1月 18, 2016
```
Fix segfault
```
  a5735a1b
17 1月, 2016 2 次提交
- Fix the cudnn context problem. · f986e0dd
  由 Arnaud Bergeron 提交于 1月 16, 2016
  
  f986e0dd
- Revert "Set the proper context when destroying the cudnn handle." · 6b7a3d16
  由 Arnaud Bergeron 提交于 1月 16, 2016
```
This reverts commit 18385728.

THis commit is broken and makes anything that uses params in the struct_code_cleanup segfault.  Will implement alternate solution.
```
  6b7a3d16
16 1月, 2016 3 次提交
- Fix grad optimization. · b1c5a67e
  由 Arnaud Bergeron 提交于 1月 11, 2016
  
  b1c5a67e
- raise a warning when the old file is imported · 91fe71d9
  由 Samira Shabanian 提交于 1月 15, 2016
  
  91fe71d9
- Merge pull request #3858 from mohammadpz/try_blas_flag · ede48744
  由 Frédéric Bastien 提交于 1月 15, 2016
```
auxiliary function 'try_blas_flags' added.
```
  ede48744
15 1月, 2016 2 次提交
- Merge pull request #3867 from nouiz/small · 7bad2e2a
  由 abergeron 提交于 1月 14, 2016
```
small stuff, tests, doc, docgen.py bugfix, doc, small code cleanup
```
  7bad2e2a
- Merge pull request #3869 from kmike/remove-scan-warning · c282d072
  由 carriepl 提交于 1月 14, 2016
```
scan: remove unconditional warning when strict=True is used
```
  c282d072