提交 · 4dd936f986e71f642fd4d7a7aa34e220eec61528 · testgroup / pytensor

04 7月, 2012 2 次提交
- Merge pull request #737 from goodfeli/fix_grad_name · 4dd936f9
  由 Razvan Pascanu 提交于 7月 04, 2012
```
fixed gradient name bug
```
  4dd936f9
- fixed gradient name bug · c7b9b3ce
  由 Ian Goodfellow 提交于 7月 03, 2012
  
  c7b9b3ce
03 7月, 2012 2 次提交
- Merge pull request #733 from bouchnic/sparse_opt · 034bb0d2
  由 nouiz 提交于 7月 03, 2012
```
Fix bug from local_csm_grad_c in mode FAST_COMPILE.
```
  034bb0d2
- Fix bug from local_csm_grad_c in mode FAST_COMPILE. · 0e299f76
  由 Nicolas Bouchard 提交于 7月 03, 2012
  
  0e299f76
30 6月, 2012 7 次提交
- Merge pull request #715 from nouiz/GpuAdvSub1 · 572bf565
  由 James Bergstra 提交于 6月 29, 2012
```
New version of GpuAdvancedSubtensor1 with gpu code.
```
  572bf565
- Added more tests. · 5dfcd8b4
  由 Frederic 提交于 6月 29, 2012
  
  5dfcd8b4
- Merge pull request #727 from bouchnic/extra_ops · 804c7e4b
  由 nouiz 提交于 6月 29, 2012
```
Add more tests and correction to RepeatOp.
```
  804c7e4b
- Add more tests. · 01be2efc
  由 Nicolas Bouchard 提交于 6月 29, 2012
  
  01be2efc
- Merge pull request #723 from bouchnic/extra_ops · 4f524015
  由 nouiz 提交于 6月 29, 2012
```
Add grad implementation for repeat for scalar.
```
  4f524015
- Make GpuAdvancedSubtensor1 use int64 for indices to make sure we support all index number. · 894d6665
  由 Frederic 提交于 6月 29, 2012
  
  894d6665
- Fix bug in RepeatOp.infer_shape and redo tests. · ba483b3f
  由 Nicolas Bouchard 提交于 6月 29, 2012
  
  ba483b3f
29 6月, 2012 10 次提交
- Merge pull request #725 from nouiz/nvcc · 15f2a18f
  由 nouiz 提交于 6月 29, 2012
```
Nvcc
```
  15f2a18f
- correctly skip cuda test when the cuda back-end is not available. · ab05b05d
  由 Frederic 提交于 6月 29, 2012
  
  ab05b05d
- correctly disable the cuda back-end when cuda_ndarray.so is there but not nvcc. · eec8002e
  由 Frederic 提交于 6月 29, 2012
  
  eec8002e
- Fix RepeatOp.grad · 63b0d7ea
  由 Nicolas Bouchard 提交于 6月 29, 2012
  
  63b0d7ea
- Merge pull request #724 from nouiz/small · 46a7640c
  由 nouiz 提交于 6月 28, 2012
```
Small
```
  46a7640c
- disable the CUDA back-end if nvcc is not there even if we can load cuda_ndarray.so · bac3c2f7
  由 Frederic 提交于 6月 28, 2012
  
  bac3c2f7
- Add GPU timming in check_blas.py · 80ded7a8
  由 Frederic 提交于 6月 28, 2012
  
  80ded7a8
- Add more info in the docstring of verify_grad. · 48b4ff5b
  由 Frederic 提交于 6月 28, 2012
  
  48b4ff5b
- Fix bug in grad for axis=-1. · f815e119
  由 Nicolas Bouchard 提交于 6月 28, 2012
  
  f815e119
- Add grad implementation for repeat for scalar. · e0ad0fd7
  由 Nicolas Bouchard 提交于 6月 28, 2012
  
  e0ad0fd7
28 6月, 2012 2 次提交
- Merge pull request #722 from dwf/acknowl_fix · e6f30b8f
  由 lamblin 提交于 6月 27, 2012
```
Minor fixes to acknowledgement.txt
```
  e6f30b8f
- Minor fixes to acknowledgement.txt · ca9a38f5
  由 David Warde-Farley 提交于 6月 27, 2012
  
  ca9a38f5
27 6月, 2012 11 次提交
- Merge pull request #681 from nouiz/conv · a9a8cb77
  由 lamblin 提交于 6月 26, 2012
```
Parallel Conv
```
  a9a8cb77
- Added a first version of an acknowledgement file. · d40f534a
  由 Frederic 提交于 6月 07, 2012
  
  d40f534a
- Add the function that compute the number of cores. · 244440a0
  由 Frederic 提交于 6月 07, 2012
```
It is taked from the pyprocessing project that is under the same license.
```
  244440a0
- Added doc on conv2d with openmp and small image/filter. · 4b9d675e
  由 Frederic 提交于 6月 07, 2012
  
  4b9d675e
- Manually disable the parallel conv when there is only possibly 1 thread running. · 7444e2b3
  由 Frederic 提交于 5月 30, 2012
  
  7444e2b3
- When only 1 core or OMP_NUM_THREADS==1, don't enable openmp by default. · aa71b758
  由 Frederic 提交于 5月 28, 2012
  
  aa71b758
- Better speed test and use the openmp conv code when the flag is enabled. · b6dc1b14
  由 Frederic 提交于 5月 24, 2012
  
  b6dc1b14
- Make a Theano flag openmp and use it in ConvOp to allow to disable it. · 04fde4e4
  由 Frederic 提交于 5月 16, 2012
```
In case it cause too much additionel overhead for small image/filter, I want to enable the user to disable it.
```
  04fde4e4
- Parallelized convolution with OpenMP. · 583a7553
  由 Frederic 提交于 5月 15, 2012
  
  583a7553
- More assertion outside of loop! · eb894aae
  由 Frederic 提交于 5月 15, 2012
  
  eb894aae
- Merge the 2 c_compile_args method in ConvOp! · b816b247
  由 Frederic 提交于 5月 15, 2012
  
  b816b247
26 6月, 2012 5 次提交
- fix on windows. · e6b2160a
  由 Frederic 提交于 6月 26, 2012
  
  e6b2160a
- remove comments. · 98df49b5
  由 Frederic 提交于 6月 26, 2012
  
  98df49b5
- Many fix following the code review. · bd00e506
  由 Frederic 提交于 6月 26, 2012
```
- add/update comments
- rename assert_fast
- fix crash
- remove useless code

This do change anything related to the int casted to float32.
```
  bd00e506
- Merge pull request #720 from nouiz/doc · 9aa36490
  由 nouiz 提交于 6月 25, 2012
```
small doc update following email from John Salvatier.

I merge it myself as this is a trivial doc change.
```
  9aa36490
- small doc update following email from John Salvatier. · 11e6029d
  由 Frederic 提交于 6月 25, 2012
  
  11e6029d
23 6月, 2012 1 次提交
- Merge pull request #712 from bouchnic/sparse_opt · ded60972
  由 nouiz 提交于 6月 22, 2012
```
fix gh-363 Move optimization from basic.py to opt.py.
```
  ded60972