 Ryan Curtin
|
6491aaebc1
|
Add a Python3 travis job to the matrix.
|
2018-01-02 14:40:24 -05:00 |
|
|
Ryan Curtin
|
c09c8fd165
|
Work around disappointing pandas versioning issue.
New versions of pandas support this constructor type; older versions do not. So
this workaround seems to solve the issue.
|
2017-12-29 19:11:50 -05:00 |
|
|
Ryan Curtin
|
c62fd68372
|
Fix incorrect dtype checks.
|
2017-12-29 19:11:25 -05:00 |
|
|
Ryan Curtin
|
5e937c90e5
|
Better strategy for string conversion---handle it after the initial conversion.
|
2017-12-29 17:17:11 -05:00 |
|
|
Ryan Curtin
|
185b559a5c
|
Handle string conversions correctly in Python 3.
|
2017-12-27 19:25:37 -05:00 |
|
|
Ryan Curtin
|
937fc5a304
|
Use relative imports explicitly instead of implicitly.
|
2017-12-25 23:07:32 -05:00 |
|
|
Marcus Edel
|
09ab63cf11
|
Merge pull request #1164 from sourabhvarshney111/master
Added Nadam optimizer implementation.
|
2017-12-21 19:07:20 +01:00 |
|
|
Ryan Curtin
|
65bd318663
|
Merge pull request #1175 from rcurtin/lrfix
Fix logistic regression tolerance checks.
|
2017-12-21 09:29:11 -05:00 |
|
|
Marcus Edel
|
a81956777b
|
Resolve merge and minor style issues.
|
2017-12-21 15:09:55 +01:00 |
|
|
Marcus Edel
|
6b7a7b072b
|
Merge branch 'master' into master
|
2017-12-21 13:54:57 +01:00 |
|
|
Ryan Curtin
|
c1d43e602a
|
Merge pull request #1170 from rcurtin/lstm-batch
Fix recurrent layers' compatibility with batch sizes > 1.
|
2017-12-20 18:47:20 -05:00 |
|
|
Ryan Curtin
|
f241d9ac13
|
Not sure how this mistake got in...
|
2017-12-20 13:27:03 -05:00 |
|
|
Ryan Curtin
|
83128ddfcc
|
The tolerance can also be 0.
|
2017-12-20 10:41:51 -05:00 |
|
|
Ryan Curtin
|
d494dd16f9
|
Update comments and apply fix to spalera_sgd.
|
2017-12-20 09:56:16 -05:00 |
|
|
Marcus Edel
|
70d98e3ed8
|
Remove 'const' type qualifier since it has no effect on the on return type.
|
2017-12-20 14:27:55 +01:00 |
|
|
Marcus Edel
|
ee5fb7db48
|
Merge pull request #1168 from haritha1313/ams_grad
Add AMSGrad optimizer implementation.
|
2017-12-20 14:04:51 +01:00 |
|
|
Haritha
|
0555e73337
|
ams_grad style edits
|
2017-12-20 11:58:26 +05:30 |
|
|
Haritha
|
7430eabac2
|
ams_grad style edits
|
2017-12-20 11:56:46 +05:30 |
|
|
Haritha
|
51e7a2a01f
|
ams_grad optimizer implementation
|
2017-12-20 11:51:40 +05:30 |
|
|
Sourabh Varshney
|
534bc65ee7
|
Added citation url
|
2017-12-19 21:33:42 +05:30 |
|
|
Ryan Curtin
|
aa83402f7e
|
Fix too-long lines.
|
2017-12-18 16:20:50 -05:00 |
|
|
Ryan Curtin
|
0e0201a5f5
|
Remove debugging output from GRU.
|
2017-12-18 14:49:28 -05:00 |
|
|
Ryan Curtin
|
735d2b97a1
|
Reset size of internal parameters whenever batch size changes.
We can call set_size() even when the batch size decreased---Armadillo will reuse
memory if we shrink the matrix size (if it can).
|
2017-12-18 14:48:17 -05:00 |
|
|
Ryan Curtin
|
45d580ddc1
|
Handle case where batchSize > maxIterations.
Don't do extra iterations!
|
2017-12-18 14:47:56 -05:00 |
|
|
Ryan Curtin
|
211346f428
|
Refactor batch size tests. This meant I needed to add RNN::Reset().
|
2017-12-18 14:47:26 -05:00 |
|
|
Sourabh Varshney
|
1aaf6d27e3
|
Added some correction in implementation
|
2017-12-17 11:31:06 +05:30 |
|
|
Sourabh Varshney
|
70efb6886f
|
Some more style changes
|
2017-12-16 18:51:12 +05:30 |
|
|
Sourabh Varshney
|
1705902e91
|
Removed Typo Error
|
2017-12-16 11:12:26 +05:30 |
|
|
Sourabh Varshney
|
ba66ef08c2
|
Removed extra whitespace
|
2017-12-16 11:04:15 +05:30 |
|
|
Sourabh Varshney
|
19ff6e8fc0
|
Made the quoted changes
|
2017-12-16 11:00:38 +05:30 |
|
|
Ryan Curtin
|
86d5898f2e
|
Merge pull request #1166 from rcurtin/haiku
Fix Haiku: link against librt
|
2017-12-14 18:55:02 -05:00 |
|
|
Haritha Sreedharan Nair
|
403f767751
|
Update adam.hpp
|
2017-12-14 20:26:08 +05:30 |
|
|
Sourabh Varshney
|
9883434c17
|
Corrected Line confliction
|
2017-12-14 17:02:39 +05:30 |
|
|
Sourabh Varshney
|
966a1503f4
|
Constructor Initialization for a parameter
|
2017-12-14 16:59:07 +05:30 |
|
|
Sourabh Varshney
|
aac0270c46
|
Corrected Typo
|
2017-12-14 16:34:33 +05:30 |
|
|
Sourabh Varshney
|
5777c6ad40
|
Applied Style Improvements
|
2017-12-14 16:23:54 +05:30 |
|
|
Sourabh Varshney
|
d2baab465b
|
Applied Style Improvements
|
2017-12-14 15:36:51 +05:30 |
|
|
Sourabh Varshney
|
e463bb03d5
|
Changed Nadam as optimizing update on Adam
|
2017-12-14 15:12:23 +05:30 |
|
|
Haritha
|
d267dd45f9
|
ams_grad optimizer
|
2017-12-14 13:49:09 +05:30 |
|
|
Ryan Curtin
|
2290fe2be7
|
Fix recurrent layers' compatibility with batch sizes > 2.
Also add tests.
|
2017-12-13 16:26:54 -05:00 |
|
|
Ryan Curtin
|
8ca93e49d0
|
&& needs another sudo. :)
|
2017-12-13 11:23:37 -05:00 |
|
|
Haritha
|
ebd0cb3f30
|
AMSGrad optimizer
|
2017-12-13 21:53:20 +05:30 |
|
|
Haritha
|
b86f7d1fcf
|
AMSGrad optimizer
|
2017-12-13 21:50:07 +05:30 |
|
|
Ryan Curtin
|
c8a97fb018
|
Another attempt to fix when apt refuses to work on Travis.
|
2017-12-13 10:33:34 -05:00 |
|
|
Sourabh Varshney
|
948516ef8e
|
Applied more style changes
|
2017-12-13 16:40:00 +05:30 |
|
|
Sourabh Varshney
|
fc45099367
|
Applied more style improvements
|
2017-12-13 16:19:25 +05:30 |
|
|
Sourabh Varshney
|
3633cd3cac
|
Applied style improvements
|
2017-12-13 16:06:38 +05:30 |
|
|
Haritha
|
28e471299b
|
ams_grad optimizer implementation
|
2017-12-13 13:08:24 +05:30 |
|
|
Zoltán Mizsei
|
6eeced182c
|
Haiku got no -lrt
|
2017-12-12 09:39:00 -05:00 |
|
|
Ryan Curtin
|
b223998e3b
|
Link to bindings tutorial.
|
2017-12-11 10:08:52 -05:00 |
|