- 20 2月, 2014 2 次提交
-
-
由 Frédéric Bastien 提交于
Small modif to allow the buildbot script to easily submit new results wi...
-
由 Frederic 提交于
-
- 19 2月, 2014 1 次提交
-
-
由 abergeron 提交于
[MRG]Better error message when a thunk fail: print user code line when node was created
-
- 18 2月, 2014 8 次提交
-
-
由 Razvan Pascanu 提交于
Use gradient_undefined instead of None in MRG uniform
-
由 Frederic Bastien 提交于
this is more clear and allow to see all of them at once.
-
由 Frédéric Bastien 提交于
Fixed memory leaks in cumsum and cumprod.
-
由 Frederic Bastien 提交于
-
由 Frederic Bastien 提交于
-
由 Marc-Alexandre Cote 提交于
-
由 Frederic Bastien 提交于
Raise the backtrace limit to caught where v1 + v2 is created in user script. Remove the add_tag_trace from the bt.
-
由 Frederic Bastien 提交于
-
- 17 2月, 2014 1 次提交
-
-
由 Frederic Bastien 提交于
-
- 15 2月, 2014 1 次提交
-
-
由 abergeron 提交于
Conversion of GpuSoftmax and GpuSoftmaxWithBias to the new backend
-
- 14 2月, 2014 7 次提交
-
-
由 abergeron 提交于
Fix unpickling of Elemwise op since openmp changes.
-
由 Frederic 提交于
-
由 Frédéric Bastien 提交于
Tag slow tests
-
由 Pierre Luc Carrier 提交于
-
由 Pierre Luc Carrier 提交于
-
由 Frédéric Bastien 提交于
Don't try to use the c_code for GpuElemwise on opencl since it's cuda-only.
-
由 Arnaud Bergeron 提交于
-
- 12 2月, 2014 2 次提交
-
-
由 Pierre Luc Carrier 提交于
Replace content of c_init_code by reference to function setup_ext_cuda() in GpuSoftmax and GpuSoftmaxWithBias
-
由 Pierre Luc Carrier 提交于
-
- 11 2月, 2014 7 次提交
-
-
由 Frédéric Bastien 提交于
Elemwise openmp
-
由 AlOa 提交于
-
由 abergeron 提交于
Fix Sparse grad crash and implement mixed dtype in sparse Mul/Add
-
由 Frederic 提交于
-
由 Frederic 提交于
-
由 Frederic 提交于
-
由 Frederic 提交于
I changed how format is specified. There was no bug as the input variable are put in the key. Now I make explicit in the op key the format. This simplify at a few places the code too.
-
- 10 2月, 2014 11 次提交