提交 · 5a0e994365c3d4f8079e2bb92df6b3c2f2a89bca · testgroup / pytensor · GitLab

30 1月, 2010 5 次提交
- corrections to DebugMode test of checks_isfinite · 5a0e9943
  由 James Bergstra 提交于 1月 29, 2010
  
  5a0e9943
- s/TypeError/ValueError in TensorType.filter when non-finite elements are not allowed · ba49f1ec
  由 James Bergstra 提交于 1月 29, 2010
  
  ba49f1ec
- added ValueError as an exception that is intercepted by Type.is_valid_value · a4481376
  由 James Bergstra 提交于 1月 29, 2010
  
  a4481376
- s/math.log/numpy.log in scalar.basic gives correct behaviour for negative values · b91d83f2
  由 James Bergstra 提交于 1月 29, 2010
  
  b91d83f2
- added try-except block to cuda.__init__ for when compilation and loading of cuda_ndarray fails · d00f619d
  由 James Bergstra 提交于 1月 29, 2010
  
  d00f619d
29 1月, 2010 22 次提交
- fix in the init of cuda. · ae613167
  由 Frederic Bastien 提交于 1月 29, 2010
  
  ae613167
- added nvcc.compiler_bindir configuration Variable · 44bf5c63
  由 James Bergstra 提交于 1月 29, 2010
  
  44bf5c63
- using_gpu doc gives example with borrow=True for better performance · d0618d6d
  由 James Bergstra 提交于 1月 29, 2010
  
  d0618d6d
- optimized TensorType.filter because all inputs to function are passed through it · 575f8edc
  由 James Bergstra 提交于 1月 29, 2010
  
  575f8edc
- Optimized Function.__call__'s assignment of positional arguments. · d66b0368
  由 James Bergstra 提交于 1月 29, 2010
  
  d66b0368
- moved print to the logger. · d92ce3f2
  由 Frederic Bastien 提交于 1月 28, 2010
  
  d92ce3f2
- check that that we don't load an old version of cuda_ndarray. If so, print a warning. · 80e91c18
  由 Frederic Bastien 提交于 1月 28, 2010
  
  80e91c18
- typo fix in last commit. Fix test following the removing of convolve2. · 4880887e
  由 Frederic Bastien 提交于 1月 28, 2010
  
  4880887e
- now record the theano fct call time and fn call time in all Mode, but don't print them. · 0e9af9c4
  由 Frederic Bastien 提交于 1月 28, 2010
```
Usefull to track the overhead of theano fct call.
```
  0e9af9c4
- Only formatting changes (wrapping lines to 80 cols), spacing etc. · 7a95162c
  由 gdesjardins 提交于 1月 28, 2010
```
Committing before actual changes take place.
```
  7a95162c
- fix the print comment. · 16d029f9
  由 Frederic Bastien 提交于 1月 28, 2010
  
  16d029f9
- move speed test fct to its own and update it to test speed all case of unroll_patch · f93b1c29
  由 Frederic Bastien 提交于 1月 28, 2010
  
  f93b1c29
- lower profiler overhead. · 304cbe3f
  由 Frederic Bastien 提交于 1月 28, 2010
  
  304cbe3f
- Update documentation to reflect right precedence rule in THEANO_FLAGS and THEANORC. · 89575768
  由 Pascal Lamblin 提交于 1月 28, 2010
  
  89575768
- merge · ecd0c486
  由 Pascal Lamblin 提交于 1月 28, 2010
  
  ecd0c486
- THEANO_FLAGS and THEANORC environment variables are now right-dominant… · 1c90389e
  由 Pascal Lamblin 提交于 1月 28, 2010
```
THEANO_FLAGS and THEANORC environment variables are now right-dominant (right-most definitions have precedence over the ones on their left)
```
  1c90389e
- fix error msg. · ea8f3186
  由 Frederic Bastien 提交于 1月 28, 2010
  
  ea8f3186
- merge · 5fba2f7c
  由 Pascal Lamblin 提交于 1月 28, 2010
  
  5fba2f7c
- remove a little bit of profiler overhead. · 8263b28e
  由 Frederic Bastien 提交于 1月 28, 2010
  
  8263b28e
- More documentation in conv.py · 01e9856a
  由 Pascal Lamblin 提交于 1月 28, 2010
  
  01e9856a
- Merged · b13d0cfd
  由 Olivier Delalleau 提交于 1月 28, 2010
  
  b13d0cfd
- Fixed bug in theano._asarray: a new view was created even when this was not necessary · 5fbd2cd7
  由 Olivier Delalleau 提交于 1月 28, 2010
  
  5fbd2cd7
28 1月, 2010 13 次提交
- Use new attributes of config · 561064b3
  由 Pascal Lamblin 提交于 1月 27, 2010
  
  561064b3
- Import theano before using theano._asarray · 5aa9743f
  由 Pascal Lamblin 提交于 1月 27, 2010
  
  5aa9743f
- revisions to tutorial/using_gpu · 7d1eb08b
  由 James Bergstra 提交于 1月 27, 2010
  
  7d1eb08b
- removed "noddy" from the CudaNdarrayType name · ee1f4acf
  由 James Bergstra 提交于 1月 27, 2010
  
  ee1f4acf
- adding link from installation file to documentation of config file · 42bd8ad1
  由 James Bergstra 提交于 1月 27, 2010
  
  42bd8ad1
- THEANORC supports multiple ":"-delimited theanorc files · 8b62d9e4
  由 James Bergstra 提交于 1月 27, 2010
  
  8b62d9e4
- removed old commented-out glossary text · aa839c64
  由 James Bergstra 提交于 1月 27, 2010
  
  aa839c64
- removed unused link from introduction · 44ed0e77
  由 James Bergstra 提交于 1月 27, 2010
  
  44ed0e77
- added tutorial page on using GPU · 1c1342e0
  由 James Bergstra 提交于 1月 27, 2010
  
  1c1342e0
- added a note to the tutorial examples about how the optimizer simplifies the gradient expression · 169a2461
  由 James Bergstra 提交于 1月 27, 2010
  
  169a2461
- added shared and __src_version__ to library doc · ea093c2c
  由 James Bergstra 提交于 1月 27, 2010
  
  ea093c2c
- Merged · 85b6de00
  由 Olivier Delalleau 提交于 1月 27, 2010
  
  85b6de00
- Replacing numpy.asarray by theano._asarray when a dtype conversion may be… · 28e2a65a
  由 Olivier Delalleau 提交于 1月 27, 2010
```
Replacing numpy.asarray by theano._asarray when a dtype conversion may be performed, to work around Numpy ticket 870 defect
```
  28e2a65a