提交 · a621068783be2d590110ae59ea0a3e25cd7c5d2a · testgroup / pytensor

22 4月, 2015 4 次提交
- [CRASH] Implement broadcasting of GpuCrossentropySoftmax1HotWithBiasDx · a6210687
  由 Frederic 提交于 4月 21, 2015
  
  a6210687
- Merge pull request #2806 from carriepl/scan_new_backend · b24d1a8d
  由 abergeron 提交于 4月 21, 2015
```
[CRASH with Scan in new backend]
```
  b24d1a8d
- Handle case where v has 0 dimensions · 2adc582c
  由 --global 提交于 4月 21, 2015
  
  2adc582c
- Merge pull request #2521 from sisp/local_useless_incsubtensor_alloc · f276201c
  由 Frédéric Bastien 提交于 4月 21, 2015
```
[WIP] New optimization: useless incsubtensor alloc
```
  f276201c
21 4月, 2015 6 次提交
- Merge pull request #2794 from MartinThoma/master · c8537817
  由 Frédéric Bastien 提交于 4月 21, 2015
```
theano/tensor/nnet/nnet.py: Wrote docstring for Softmax
```
  c8537817
- Merge pull request #2769 from carriepl/equal_computations · a8df83f8
  由 Pascal Lamblin 提交于 4月 20, 2015
```
Remove dependance on io_toposort in equal_computations()
```
  a8df83f8
- Merge pull request #2761 from abergeron/pep8 · 2b892bc2
  由 Frédéric Bastien 提交于 4月 20, 2015
```
A bunch of pep8 work
```
  2b892bc2
- Merge pull request #2780 from ejls/master · 09944cfe
  由 Xavier Bouthillier 提交于 4月 20, 2015
```
Add an optimization that remove Split with only 1 split. (Close #2545)
```
  09944cfe
- Fix (fatal) nit that slipped in with the rebase. · ec103106
  由 Arnaud Bergeron 提交于 4月 20, 2015
  
  ec103106
- Merge pull request #2798 from nouiz/tests · f49eadd2
  由 Frédéric Bastien 提交于 4月 20, 2015
```
Tests
```
  f49eadd2
20 4月, 2015 6 次提交
- Fix test in daily buildbot. It do not work with Theano flags vm.lazy=True · 64f15818
  由 Frederic 提交于 4月 20, 2015
  
  64f15818
- Fix test with floatX=float32 · ed0b4e5f
  由 Frederic 提交于 4月 20, 2015
  
  ed0b4e5f
- Fix daily build test due to rounding problems · 301c4a23
  由 Frederic 提交于 4月 20, 2015
  
  301c4a23
- Move test to where it should · 41075539
  由 Frederic 提交于 4月 20, 2015
  
  41075539
- test pep8 and better error msg · 5ae5123d
  由 Frederic 提交于 4月 20, 2015
  
  5ae5123d
- fix test with floatX=float32 · ca732b67
  由 Frederic 提交于 4月 20, 2015
  
  ca732b67
18 4月, 2015 3 次提交
- theano/tensor/nnet/nnet.py: Wrote docstring for Softmax · 66b6d812
  由 Martin Thoma 提交于 4月 18, 2015
  
  66b6d812
- Merge pull request #2786 from nouiz/cache · 52a98808
  由 Pascal Lamblin 提交于 4月 17, 2015
```
Cache
```
  52a98808
- Add an optimization that remove Split with only 1 split. (Close #2545) · 9ce29e17
  由 Étienne Simon 提交于 4月 16, 2015
  
  9ce29e17
17 4月, 2015 21 次提交
- Revert change that doesn't pass test_flake8 after 2to3. · 7315d4fe
  由 Arnaud Bergeron 提交于 4月 17, 2015
  
  7315d4fe
- Make cuda.py importable even if cuda is not available. · 5ce861c1
  由 Arnaud Bergeron 提交于 4月 17, 2015
  
  5ce861c1
- Fix small remaining issues. · 0ea2c730
  由 Arnaud Bergeron 提交于 4月 13, 2015
  
  0ea2c730
- Remove more support for python < 2.6 · b4181fcb
  由 Arnaud Bergeron 提交于 4月 13, 2015
  
  b4181fcb
- Make test_flake8 pass itself. · 7540e33a
  由 Arnaud Bergeron 提交于 4月 13, 2015
  
  7540e33a
- Make raise_op pass flake8 + spread the gospel of __props__ · 4f3e520f
  由 Arnaud Bergeron 提交于 4月 13, 2015
  
  4f3e520f
- Make printing.py pass test_flake8 · 66dc1578
  由 Arnaud Bergeron 提交于 4月 13, 2015
  
  66dc1578
- Make theano/compat/* pass the flake8 test. · 837f6ad6
  由 Arnaud Bergeron 提交于 4月 13, 2015
  
  837f6ad6
- Remove support for python < 2.6. · 856a3662
  由 Arnaud Bergeron 提交于 4月 13, 2015
  
  856a3662
- Stop importing all and any from compat as they are builtins since 2.5 · 347a46eb
  由 Arnaud Bergeron 提交于 4月 13, 2015
  
  347a46eb
- Move DefaultOrderedDict to a more natural location. · 468fe6d2
  由 Arnaud Bergeron 提交于 4月 13, 2015
  
  468fe6d2
- Make flake8 stop complaining about __init__.py (which will keep having errors). · cd0b520b
  由 Arnaud Bergeron 提交于 4月 13, 2015
  
  cd0b520b
- Merge pull request #2245 from lamblin/clip_grad_doc · 5ef9fec9
  由 Frédéric Bastien 提交于 4月 17, 2015
```
Update documentation for gradient of clip
```
  5ef9fec9
- update URL · a05f077b
  由 Frederic 提交于 4月 17, 2015
  
  a05f077b
- get the lock if we err too frequently. · 290620b2
  由 Frederic 提交于 4月 17, 2015
  
  290620b2
- Move tests to theano/tensor, call verify_grad · 41518f9e
  由 Pascal Lamblin 提交于 11月 06, 2014
```
Calling theano.grad on the graph built from theano.scalar.clip
is not working, so we have to use theano.tensor.clip instead.
```
  41518f9e
- Update documentation for gradient of `clip` · 7ef9e747
  由 Pascal Lamblin 提交于 11月 06, 2014
  
  7ef9e747
- Merge pull request #2773 from nouiz/fix_nan · ba5f2a3a
  由 Pascal Lamblin 提交于 4月 16, 2015
```
[REGRESSION] Fix nan 
```
  ba5f2a3a
- Update test following code review and fix DebugMode · 78075be5
  由 Frederic Bastien 提交于 4月 16, 2015
  
  78075be5
- Merge pull request #2777 from carriepl/scan_new_backend · 9faa62d7
  由 abergeron 提交于 4月 16, 2015
```
Add missing arguments to GpuKernel_init
```
  9faa62d7
- Add missing arguments to GpuKernel_init · 286b16ef
  由 --global 提交于 4月 16, 2015
  
  286b16ef