提交 · 1cd49b157dc2e78f30188af3faf2ccbdb4f50708 · testgroup / pytensor

04 6月, 2014 4 次提交
- rm print in test · 1cd49b15
  由 Frederic 提交于 5月 30, 2014
  
  1cd49b15
- small fix. · 09a441e9
  由 Frederic 提交于 5月 30, 2014
  
  09a441e9
- pep8 · 886dc3ad
  由 Frederic 提交于 5月 30, 2014
  
  886dc3ad
- Add GpuSqrSumAx0 to lower the memory usage on the GPU. · 43d45788
  由 Frederic 提交于 5月 30, 2014
  
  43d45788
02 6月, 2014 2 次提交
- Merge pull request #1893 from nouiz/import · 73d8d175
  由 Frédéric Bastien 提交于 6月 02, 2014
```
Allow import without the gpu back-end.
```
  73d8d175
- Allow import without the gpu back-end. · b98e5b25
  由 Frederic 提交于 6月 02, 2014
```
needed for docgen.
```
  b98e5b25
31 5月, 2014 2 次提交
- Merge pull request #1889 from AlassaneNdiaye/fix_test_cross_validation_bug · 38e29b17
  由 Frédéric Bastien 提交于 5月 30, 2014
```
Fixed bug detected by test_cross_validation.py:test_train_cv
```
  38e29b17
- Fixed bug where the variable name was defined but not used. · 26d715ff
  由 Alassane Ndiaye 提交于 5月 30, 2014
  
  26d715ff
30 5月, 2014 5 次提交
- Merge pull request #1887 from vdumoulin/rpi_bugfix · 9625521b
  由 Frédéric Bastien 提交于 5月 30, 2014
```
Fix Raspberry Pi-related bug
```
  9625521b
- Fix wrong logic in ARM architecture detection for compilation · b4289338
  由 Vincent Dumoulin 提交于 5月 29, 2014
  
  b4289338
- Merge pull request #1886 from nouiz/fftconv_import · c60b7acb
  由 Frédéric Bastien 提交于 5月 29, 2014
```
[BUG] fix import theano.sandbox.cuda work when there is no pycuda or scikits.cuda
```
  c60b7acb
- make theano.sandbox.cuda import work when there is no pycuda or scikits.cuda · 27135a7d
  由 Frederic 提交于 5月 29, 2014
  
  27135a7d
- Merge pull request #1882 from abergeron/cuda_fftconv · b7bcc91a
  由 Frédéric Bastien 提交于 5月 29, 2014
```
Cuda fftconv assert
```
  b7bcc91a
29 5月, 2014 3 次提交
- Don't use ==, use T.eq. · 1a60236f
  由 Arnaud Bergeron 提交于 5月 28, 2014
  
  1a60236f
- Add assert in the conv2d_fft graph. · 3b075b19
  由 Arnaud Bergeron 提交于 5月 28, 2014
  
  3b075b19
- Add support for custom message in Assert. · 6aa4a034
  由 Arnaud Bergeron 提交于 5月 28, 2014
  
  6aa4a034
28 5月, 2014 7 次提交
- Merge pull request #1878 from abergeron/gpuarray_doc · ba81e61d
  由 Frédéric Bastien 提交于 5月 28, 2014
```
Gpuarray doc
```
  ba81e61d
- Merge pull request #1870 from abergeron/cuda_fftconv · 75550055
  由 Frédéric Bastien 提交于 5月 27, 2014
```
Cuda fftconv
```
  75550055
- Finally figured out the header problem. Now it should be what we want. · eee7cdd8
  由 Arnaud Bergeron 提交于 5月 27, 2014
  
  eee7cdd8
- Remove the "Full speed" section and link to the borrow tutorial. · a66ad5db
  由 Arnaud Bergeron 提交于 5月 27, 2014
```
Also add an example of the GPU speedup for borrow to that tutorial.
```
  a66ad5db
- Add numpy "casting" example to the old backend section. · 70543486
  由 Arnaud Bergeron 提交于 5月 27, 2014
  
  70543486
- Merge pull request #1874 from abergeron/gpuarray_nvcc · 5f219fca
  由 Frédéric Bastien 提交于 5月 27, 2014
```
Make a proxy NVCC_compiler that does not create a second context on the card as a side effect.
```
  5f219fca
- Some fixes following comments. · bc7e4495
  由 Arnaud Bergeron 提交于 5月 27, 2014
  
  bc7e4495
27 5月, 2014 11 次提交
- Merge pull request #1876 from jlowin/disconnected_inputs · 8477307a
  由 Frédéric Bastien 提交于 5月 27, 2014
```
Fix: disconnected_inputs arg not getting passed from jacobian to grad
```
  8477307a
- np -> numpy · b2351b96
  由 Jeremiah Lowin 提交于 5月 26, 2014
  
  b2351b96
- Fix the device='cuda' case. · 4a98bd66
  由 Arnaud Bergeron 提交于 5月 26, 2014
  
  4a98bd66
- Add a section about GpuArray in the Using GPU section of the tutorial. · 3c9d6446
  由 Arnaud Bergeron 提交于 5月 26, 2014
  
  3c9d6446
- pass consider_constant to grad · 21620c33
  由 Jeremiah Lowin 提交于 5月 26, 2014
  
  21620c33
- add test for Jacobian disconnected inputs · d047b81a
  由 Jeremiah Lowin 提交于 5月 26, 2014
  
  d047b81a
- Remember the device init_dev() was called with. · a08bba5b
  由 Arnaud Bergeron 提交于 5月 26, 2014
  
  a08bba5b
- add disconnected_inputs to grad call · 1ef58767
  由 Jeremiah Lowin 提交于 5月 26, 2014
  
  1ef58767
- Make a proxy NVCC_compiler that does not create a second context on the card as a side effect. · 017b6e3e
  由 Arnaud Bergeron 提交于 3月 04, 2014
```
Cuda docs suggest that more than one context on a card may slow down operations, so this might give us a slight win.
```
  017b6e3e
- Add a --cache option to docgen to make it not disable the doctree cache. · 5b1c0b3b
  由 Arnaud Bergeron 提交于 5月 26, 2014
```
This helps immensly when writing or editing documenation by massively
speeding up documentation rebuilds (as in 2-3s versus 60-90s before).
```
  5b1c0b3b
- Fix most of the errors/warnings in the current docs. · c01948b6
  由 Arnaud Bergeron 提交于 5月 26, 2014
```
Also sprinkle a bit of pep8.
```
  c01948b6
26 5月, 2014 3 次提交
- Merge pull request #1873 from nouiz/doc · 3810c977
  由 abergeron 提交于 5月 26, 2014
```
Add doc for gamma, gammaln, psi and chi2sf.
```
  3810c977
- Merge pull request #1843 from bartvm/scan_error_message · 454df052
  由 Frédéric Bastien 提交于 5月 26, 2014
```
Fixed error message for scan op dimension mismatch
```
  454df052
- Add doc for gamma, gammaln, psi and chi2sf. · b7445e97
  由 Frederic 提交于 5月 26, 2014
  
  b7445e97
24 5月, 2014 2 次提交
- Clearer error message · aecfb7d0
  由 Bart van Merriënboer 提交于 5月 23, 2014
  
  aecfb7d0
- Merge pull request #1862 from Hengjean/TypedList · edbd70cd
  由 abergeron 提交于 5月 23, 2014
```
Implemented basic typed list with some basic op
```
  edbd70cd
23 5月, 2014 1 次提交
- Add a mention of conv2d_fft in the docs and write a better docstring for it. · 4cf06d2b
  由 Arnaud Bergeron 提交于 5月 23, 2014
  
  4cf06d2b