- 07 1月, 2012 3 次提交
-
-
由 Razvan Pascanu 提交于
added ability to use debug_assert without a message argument
-
由 Razvan Pascanu 提交于
fixed a bug where get_debug_values returns tuple(val) rather than val in
-
由 Ian Goodfellow 提交于
the case where it is fetching the value of only one variable
-
- 06 1月, 2012 3 次提交
-
-
由 Ian Goodfellow 提交于
-
由 James Bergstra 提交于
fixed error handling and doc for ifelse
-
由 Ian Goodfellow 提交于
(there was a problem that would reject a pair of branches where one was shared and the other was not, regardless of whether they had the same type)
-
- 22 12月, 2011 11 次提交
-
-
由 nouiz 提交于
Minor stuff
-
由 Olivier Delalleau 提交于
-
由 Olivier Delalleau 提交于
-
由 Olivier Delalleau 提交于
-
由 Olivier Delalleau 提交于
Updated NEWS.txt following scan change.
-
由 nouiz 提交于
Fixed some tests
-
由 Frederic 提交于
-
由 nouiz 提交于
Temporary fix to a known bug for scan.
-
由 nouiz 提交于
Added warning about sigmoid identification bug
-
由 Razvan Pascanu 提交于
Fix scan crash in recent change. The code seems fine.,
-
由 Razvan Pascanu 提交于
-
- 21 12月, 2011 1 次提交
-
-
由 Frederic 提交于
-
- 20 12月, 2011 4 次提交
-
-
由 Olivier Delalleau 提交于
-
由 Olivier Delalleau 提交于
Also updated NEWS.txt accordingly. This fixes gh-302.
-
由 Olivier Delalleau 提交于
-
由 Olivier Delalleau 提交于
Test sigm opt
-
- 18 12月, 2011 7 次提交
-
-
由 Frederic 提交于
-
由 Frederic 提交于
This fix the gpu scan optimization that was still passing an Apply as in the past.
-
由 Frederic Bastien 提交于
-
由 Frederic Bastien 提交于
-
由 Frederic Bastien 提交于
-
由 nouiz 提交于
Fixed bug in identification of sigmoid
-
由 Olivier Delalleau 提交于
It used to think that (any_constant + exp(x)) was equal to (1 + exp(x)).
-
- 14 12月, 2011 11 次提交
-
-
由 Olivier Delalleau 提交于
Fix pycuda test
-
由 nouiz 提交于
Fixed tests when on_shape_error = 'raise'
-
由 nouiz 提交于
Scan check
-
由 Razvan Pascanu 提交于
-
由 Frederic Bastien 提交于
-
由 Olivier Delalleau 提交于
It is expected that some implementations of 'infer_shape' raise a ShapeError to indicate that they do not know how to compute the shape. In such a situation, we should silently fall back to the default implementation, regardless of the value of 'on_shape_error' (whose name may be misleading for developers, since we want to catch all exceptions *except* ShapeError). Also, when a bad shape was inferred, we should not raise a ShapeError (which may be caught silently), but rather something that indicates that there is a serious bug somewhere, like an AssertionError.
-
由 nouiz 提交于
Minor typo fix and more consistent use of rand
-
由 Frederic 提交于
-
由 Razvan Pascanu 提交于
-
由 Razvan Pascanu 提交于
inputs.
-
由 Razvan Pascanu 提交于
-