提交 · b7b95ee2c8307f7d0505cd56b837d45cff2f0d35 · testgroup / pytensor

31 3月, 2012 12 次提交
- small fix following code review. · b7b95ee2
  由 Frederic 提交于 3月 30, 2012
  
  b7b95ee2
- Update NEWS.txt · be9826d9
  由 Frederic 提交于 3月 30, 2012
  
  be9826d9
- In debugmode, in the check out output memory, transfer less data to the gpu. · ef968c65
  由 Frederic 提交于 3月 30, 2012
  
  ef968c65
- allow GpuAdvancedIncSubtensor1 to broadcast automatically as the code support it. · bf797056
  由 Frederic 提交于 3月 30, 2012
  
  bf797056
- Make GpuAdvancedIncSubtensor1 work for set_subtensor. · 6e0f1fe4
  由 Frederic 提交于 3月 30, 2012
  
  6e0f1fe4
- We add dimensions to CudaNdarray to automatically broadcast before copy. · 3de42af7
  由 Frederic 提交于 3月 30, 2012
```
This affect all the function that call this inner function.
```
  3de42af7
- update test for CudaNdarray__setitem · 697d3eb7
  由 Frederic 提交于 3月 30, 2012
  
  697d3eb7
- make CudaNdarray_set_device_data const input(was needed by a previous/following commit) · 83ec4038
  由 Frederic 提交于 3月 30, 2012
  
  83ec4038
- CudaNdarray_setitem now accept more case when the new value is on the cpu. · 330a038c
  由 Frederic 提交于 3月 30, 2012
  
  330a038c
- Allow CudaNdarray_inplace_elemwise on empty array for broadcasted constant as numpy. · 5fd2cb21
  由 Frederic 提交于 3月 30, 2012
  
  5fd2cb21
- CudaNdarray_inplace_elemwise, accept input then CudaNdarray. · 19da7c2d
  由 Frederic 提交于 3月 30, 2012
  
  19da7c2d
- 80 character fix. · b9bc0142
  由 Frederic 提交于 3月 30, 2012
  
  b9bc0142
30 3月, 2012 10 次提交
- Update NEWS.txt · 241f4ffb
  由 Frederic 提交于 3月 30, 2012
  
  241f4ffb
- not verbose by default. · a67a952f
  由 Frederic 提交于 3月 30, 2012
  
  a67a952f
- indentattion · 10f865ce
  由 Frederic 提交于 3月 30, 2012
  
  10f865ce
- Make the CudaNdarray += CudaNdarray automatically add broadcasted dimensions. · 738eae2e
  由 Frederic 提交于 3月 30, 2012
```
This is needed for GpuAdvancedIncSubetensor1
```
  738eae2e
- Added fct to test compiled function. · ee76025c
  由 Frederic 提交于 3月 30, 2012
  
  ee76025c
- Update test T_subtensor.test_advanced_inc_and_set to work correctly. · c8615bdc
  由 Frederic 提交于 3月 29, 2012
  
  c8615bdc
- Add the missing definition of the theano flags · 2c8c27d3
  由 Frederic 提交于 3月 29, 2012
  
  2c8c27d3
- Update NEWS.txt file. · 1507c85a
  由 Frederic 提交于 3月 29, 2012
  
  1507c85a
- Add the commit id and date to warning msg. Also fix stacklevel. · d022c7b0
  由 Frederic 提交于 3月 29, 2012
  
  d022c7b0
- Don't apply InputToGpuOptimizer to input that are already moved to the gpu. · f4779cbb
  由 Frederic 提交于 3月 29, 2012
  
  f4779cbb
29 3月, 2012 3 次提交
- Disable GpuAdvancedIncSubtensor1 when set_instead_of_inc is True and warn about it. · 6872ae43
  由 Frederic 提交于 3月 29, 2012
  
  6872ae43
- small code refactoring. Also, there is no FAST_COMPILE on the gpu. · fbb70262
  由 Frederic 提交于 3月 28, 2012
  
  fbb70262
- make test execute more often on the GPU. · cbceb887
  由 Frederic 提交于 3月 28, 2012
  
  cbceb887
28 3月, 2012 3 次提交
- Merge pull request #543 from pascanur/fix_gradient · de717450
  由 nouiz 提交于 3月 28, 2012
```
Do not wrap in as_tensor_variable unless needed
```
  de717450
- Merge pull request #571 from lamblin/test_div_grad · 1422638a
  由 nouiz 提交于 3月 28, 2012
```
Do not test div on denominator too close to 0.
```
  1422638a
- Do not test div on denominator too close to 0. · edd21a1d
  由 Pascal Lamblin 提交于 3月 28, 2012
```
The numeric estimation of the gradient was sometimes quite off.
```
  edd21a1d
27 3月, 2012 8 次提交
- Merge pull request #570 from nouiz/fix · 2d2923f1
  由 nouiz 提交于 3月 27, 2012
```
Remove new bugged version of GpuSum commited in a pep8 commits.
```
  2d2923f1
- Remove new bugged version of GpuSum commited in a pep8 commits. · f22bac31
  由 Frederic 提交于 3月 27, 2012
  
  f22bac31
- Merge pull request #560 from nouiz/crash_fix_gpu_reshape · 568bd526
  由 lamblin 提交于 3月 26, 2012
```
Crash fix gpu reshape
```
  568bd526
- Merge pull request #569 from nouiz/blas · 0b8c20d7
  由 Olivier Delalleau 提交于 3月 26, 2012
```
Blas
```
  0b8c20d7
- fix test when config.blas.ldflags is empty. · 6e39a60b
  由 Frederic 提交于 3月 26, 2012
  
  6e39a60b
- pep8 · 76f2757a
  由 Frederic 提交于 3月 26, 2012
  
  76f2757a
- pep8 · 0c0059e9
  由 Frederic 提交于 3月 26, 2012
  
  0c0059e9
- Merge pull request #568 from ynd/sp_sandbox · 3c9177df
  由 nouiz 提交于 3月 26, 2012
```
Remove AddSSData Optimization
```
  3c9177df
26 3月, 2012 1 次提交
- remove error prone optimization · 60027c57
  由 Yann N. Dauphin 提交于 3月 26, 2012
  
  60027c57
25 3月, 2012 3 次提交
- Merge pull request #567 from lamblin/fix_scan_memory_bug · ebf5ff54
  由 Razvan Pascanu 提交于 3月 24, 2012
```
Fix elusive output-corruption bug in Scan.
```
  ebf5ff54
- Fix elusive memory-corruption bug in Scan. · e6777e0b
  由 Pascal Lamblin 提交于 3月 24, 2012
```
The problem was that when copying a part of an array onto an overlapping
part of the same array, sometimes elements were overwritten before being
copied, leading to silent data corruption.
```
  e6777e0b
- Merge pull request #566 from lamblin/fix_python24 · efade648
  由 Olivier Delalleau 提交于 3月 24, 2012
```
Import 'any' for python 2.4
```
  efade648