提交 · a9cb097cfffe001f3ca3b36da5a8384fce3d7303 · testgroup / pytensor

26 3月, 2011 12 次提交
- prevent NameError when a prior error (unusual) caused variable not to be defined · a9cb097c
  由 Josh Bleecher Snyder 提交于 3月 25, 2011
  
  a9cb097c
- Fix another test in float32. · e68b5055
  由 Pascal Lamblin 提交于 3月 25, 2011
  
  e68b5055
- Fix test in float32 · 0a18d7da
  由 Pascal Lamblin 提交于 3月 25, 2011
  
  0a18d7da
- Fix bug recently introduced in ShapeFeature · b155c92f
  由 Pascal Lamblin 提交于 3月 25, 2011
  
  b155c92f
- reword metaoptimize link · d1b923fd
  由 James Bergstra 提交于 3月 25, 2011
  
  d1b923fd
- added metaoptimize link to webpage · 976f59b7
  由 James Bergstra 提交于 3月 25, 2011
  
  976f59b7
- try to make pycuda work with gpu in exclusive mode. · 437d311b
  由 Frederic Bastien 提交于 3月 25, 2011
  
  437d311b
- 'New mechanism to create a CudaNdarray from a pycuda gpu memory and example/test for it.' · c33d5f99
  由 Frederic Bastien 提交于 3月 25, 2011
  
  c33d5f99
- 'use a new way to init pycuda to work correctly with newer version of Theano and Pycuda.' · 63b17b09
  由 Frederic Bastien 提交于 3月 25, 2011
  
  63b17b09
- Allow binary bit-wise op on int16(previous commit was for unary bit-wise) and… · f608b307
  由 Frederic Bastien 提交于 3月 25, 2011
```
Allow binary bit-wise op on int16(previous commit was for unary bit-wise) and add c code for bit-wise and, or, xor and invert.
```
  f608b307
- Allow bit-wise op to be of type int16 and add test for all allowed type. · 3912f005
  由 Frederic Bastien 提交于 3月 25, 2011
  
  3912f005
- fix comment. · 163a00fd
  由 Frederic Bastien 提交于 3月 25, 2011
  
  163a00fd
25 3月, 2011 18 次提交
- moved greedy_local_optimizer to theano/got/opt.py and renamed it to pre_greedy_local_optimizer. · 9eb0b187
  由 Frederic Bastien 提交于 3月 25, 2011
  
  9eb0b187
- pre apply more opt. · 34123842
  由 Frederic Bastien 提交于 3月 25, 2011
  
  34123842
- pre merge constant. · 692d173b
  由 Frederic Bastien 提交于 3月 25, 2011
  
  692d173b
- clarified comment. · a614fa0c
  由 Frederic Bastien 提交于 3月 25, 2011
  
  a614fa0c
- simplified code a little bit. · 07b499b1
  由 Frederic Bastien 提交于 3月 25, 2011
  
  07b499b1
- fix comment. · 3b9fbee8
  由 Frederic Bastien 提交于 3月 25, 2011
  
  3b9fbee8
- Update test of mlp, as two more cases are now working · a79a4db0
  由 Pascal Lamblin 提交于 3月 24, 2011
  
  a79a4db0
- Add more shape-tracking possibilities to ShapeFeature · 2813506a
  由 Pascal Lamblin 提交于 3月 24, 2011
  
  2813506a
- Fix a stability optimization in nnet. · f95dc2d5
  由 Pascal Lamblin 提交于 3月 24, 2011
```
The special casting case is unneeded, and makes things ugly
when infer_shape of subtensor is implemented.
```
  f95dc2d5
- Enables printing of an Env in debugprint · 7d9b8e83
  由 Pascal Lamblin 提交于 3月 24, 2011
  
  7d9b8e83
- small fix. · 7229a93e
  由 Frederic Bastien 提交于 3月 24, 2011
  
  7229a93e
- pre apply another optimization. · f0d000ac
  由 Frederic Bastien 提交于 3月 24, 2011
  
  f0d000ac
- 'Added a file to scan the compilation directory and give stats about duplicate c code compiled.' · 328ccd59
  由 Frederic Bastien 提交于 3月 24, 2011
  
  328ccd59
- removed printing to stderr in one test that was cluttering the output of nosetests · 700f0f89
  由 Razvan Pascanu 提交于 3月 24, 2011
  
  700f0f89
- Bug fix in the changeset 5532 added by Fred. · 6ba76bb3
  由 Razvan Pascanu 提交于 3月 24, 2011
  
  6ba76bb3
- Don't pre-apply optimization to node in the env. · c5dbc0d7
  由 Frederic Bastien 提交于 3月 24, 2011
  
  c5dbc0d7
- make pydotprint_variables interface more like pydotprint. · 0bf76bf5
  由 Frederic Bastien 提交于 3月 24, 2011
  
  0bf76bf5
- fix pydotprint_variables. · c49edd05
  由 Frederic Bastien 提交于 3月 24, 2011
  
  c49edd05
24 3月, 2011 10 次提交
- merge · 301305e8
  由 Razvan Pascanu 提交于 3月 24, 2011
  
  301305e8
- Reduced the number of tests which was prohibitve ( it took 2h in debug mode) · ed0a982d
  由 Razvan Pascanu 提交于 3月 24, 2011
  
  ed0a982d
- A greedy local way of applying constant folding and removing switches with · 62d8c5ea
  由 Razvan Pascanu 提交于 3月 24, 2011
```
constant conditions.
```
  62d8c5ea
- remove not used import. · bb0baa9d
  由 Frederic Bastien 提交于 3月 24, 2011
  
  bb0baa9d
- remove not used import. · 51d90e8d
  由 Frederic Bastien 提交于 3月 24, 2011
  
  51d90e8d
- use the right dtype as we don't silently downcast anymore · 711f5cfd
  由 Frederic Bastien 提交于 3月 23, 2011
  
  711f5cfd
- merge · e8f198a4
  由 Razvan Pascanu 提交于 3月 23, 2011
  
  e8f198a4
- Apperently working version of general subtensor subtensor merge. · ce827d20
  由 Razvan Pascanu 提交于 3月 23, 2011
  
  ce827d20
- Use self.dtype instead of hardcoded type in T_subtensor, so it is usable on GPU · 2d42dc14
  由 Pascal Lamblin 提交于 3月 23, 2011
  
  2d42dc14
- Correctly cast random values to floatX in some tests. · b84a6506
  由 Pascal Lamblin 提交于 3月 23, 2011
  
  b84a6506