Sam Gross
76ae03d5f1
Operate on Variables in torch.nn.init ( #4964 )
...
Once Variable and Tensor are merged the existing Variable test would
cause an infinite recursion. Instead, modify the Variables directly
inside a `no_grad()` block.
2018-02-05 11:34:05 -05:00
Soumith Chintala
99068d2e52
fix nn.init.constant example
2017-12-29 19:14:53 +09:00
Alykhan Tejani
7752fe5d4e
remove zero padding in orthogonal initialization
2017-09-14 23:13:43 -04:00
Soumith Chintala
e4c0af8b56
revert #2708 modify orthogonal init for rows<cols case
2017-09-13 18:23:43 -04:00
nadavbh12
d01adcbe0e
modify orthogonal init
2017-09-13 16:54:37 -04:00
Alykhan Tejani
f0f7b39650
fix example in docs for nn.init.calculate_gain ( #2600 )
2017-09-02 19:23:25 -04:00
Surag Nair
8b42308f71
Bug in line 381 (sparse) ( #2130 )
...
The function iterates over columns and sets "sparsity" fraction of entires in each column to 0. The number of zeros in a column (num_zeros) is then ceil(rows*sparsity)
2017-07-18 22:55:06 -04:00
Leonid Vlasenkov
46a868dab7
[Ready] Limit docs line length ( #1900 )
...
* some docs are ready
* docs
* docs
* fix some more
* fix some more
2017-07-10 10:24:54 -04:00
Francisco Massa
6626881e7a
Add Alpha Dropout ( #1775 )
2017-06-13 00:39:49 +02:00
GBLin5566
e50c7daaf9
Use Qr factorization to get orthogonal matrix in orthogonal init ( #1453 )
2017-05-04 07:11:59 -04:00
Kai Arulkumaran
48a7869b23
Doc fixes ( #1409 )
2017-04-30 08:28:19 -04:00
Kai Arulkumaran
cbb9f08b71
Add new init methods gain, eye and dirac ( #1172 )
2017-04-28 17:16:40 -04:00
Alykhan Tejani
be6322e4b5
Update nn.init docstrings to correctly reference the module ( #1001 )
2017-03-15 11:17:59 -04:00
Alykhan Tejani
01650ac9de
add torch.nn.init docs to the source folder ( #979 )
2017-03-11 10:11:30 -05:00
Alykhan Tejani
37e05485d9
added initialization schemes in torch.nn.init ( #833 )
2017-03-01 19:34:13 +01:00
Adam Paszke
63893c3fa2
Fix auto-gpu semantics for indexing
2017-01-22 18:02:40 -05:00
