 Ryan Curtin
|
55d15b16fd
|
Patch from @marcespie: no need for bash, don't gzip.
|
2018-10-09 09:21:09 -04:00 |
|
|
Ryan Curtin
|
7d0cf18463
|
clock_gettime() isn't used anymore.
|
2018-10-09 09:00:37 -04:00 |
|
|
Shikhar Jaiswal
|
eee326bf9e
|
Merge pull request #1508 from zoq/layer_serialization
Layer serialization
|
2018-09-29 04:27:12 +05:30 |
|
|
Marcus Edel
|
8f983f1f70
|
Remove default size parameter.
|
2018-09-28 23:15:18 +02:00 |
|
|
Marcus Edel
|
7f4967f4cf
|
Adjust LayerNormSerializationTest test case according to #1512.
|
2018-09-26 00:04:23 +02:00 |
|
|
Marcus Edel
|
7176589672
|
Minor style fix (remove extra space).
|
2018-09-22 14:04:15 +02:00 |
|
|
Marcus Edel
|
23f77b86d7
|
Merge branch 'master' into layer_serialization
|
2018-09-22 13:58:00 +02:00 |
|
|
Marcus Edel
|
d89678862a
|
Merge pull request #1512 from ShikharJ/LayerNorm
Correct Scaling Parameters in LayerNorm.
|
2018-09-21 13:01:44 +02:00 |
|
|
Shikhar Jaiswal
|
c54be74524
|
Correct Scaling Parameters
|
2018-09-21 00:49:56 +05:30 |
|
|
Marcus Edel
|
225f114703
|
Merge pull request #1507 from Wenhao-H/cf-comments
Correct and add examples on how to use the policies in CF.
|
2018-09-17 00:30:37 +02:00 |
|
|
Marcus Edel
|
fac6965de1
|
Fix ambiguous call to overloaded function (size).
|
2018-09-17 00:29:09 +02:00 |
|
|
Marcus Edel
|
90d2a6c1a7
|
Add a BatchNorm and LayerNorm serialization test.
|
2018-09-16 18:42:10 +02:00 |
|
|
Marcus Edel
|
b0296ceb04
|
Do not override layer parameter when loading.
|
2018-09-16 18:39:33 +02:00 |
|
|
Marcus Edel
|
2365974964
|
Use Welford method to compute the sample mean and variance.
|
2018-09-16 18:38:22 +02:00 |
|
|
Wenhao-H
|
d877034656
|
add examples in cf
|
2018-09-14 17:51:54 -04:00 |
|
|
Wenhao-H
|
f337ec2cc3
|
correct examples in code comments
|
2018-09-14 17:28:04 -04:00 |
|
|
Marcus Edel
|
cf258d1c87
|
Merge pull request #1503 from sshkhr/rl
Added reward clipping wrapper.
|
2018-09-13 13:24:44 +02:00 |
|
|
Marcus Edel
|
024e7ea794
|
Merge pull request #1495 from akhandait/concatenate_layer
[WIP] Add Concatenate layer.
|
2018-09-07 18:42:42 +03:00 |
|
|
Shashank Shekhar
|
2bf222b67c
|
Made requested changes and changed Sample method to re-use previously defined one
|
2018-09-07 10:36:03 +05:30 |
|
|
Wenhao Huang
|
bcaf2d0797
|
Merge pull request #1458 from Wenhao-H/cf-decomposition
Refactor CF DecompositionPolicy and add BiasSVD, SVD++
|
2018-09-06 12:22:53 -04:00 |
|
|
Ryan Curtin
|
d21cb336f8
|
Merge pull request #1485 from gmanlan/win-msi-installer
Windows WiX Installer
|
2018-09-05 21:30:12 -04:00 |
|
|
Ryan Curtin
|
80e6ca853f
|
Merge pull request #1500 from rcurtin/lr_speedup
Replace linear regression implementation with something faster.
|
2018-09-04 13:52:02 -04:00 |
|
|
Shashank Shekhar
|
4ec85b4186
|
Added test for Q-Learning convergence with reward clipping
|
2018-09-04 21:59:50 +05:30 |
|
|
Wenhao-H
|
5f0310714a
|
various minor fixes
|
2018-09-04 01:29:30 -04:00 |
|
|
Shashank Shekhar
|
dd52cf3225
|
Fix naming error
|
2018-09-04 04:25:52 +05:30 |
|
|
Shashank Shekhar
|
ad2fc20cd3
|
Added sampling with next state parameter for interacting with q learning function
|
2018-09-04 03:00:14 +05:30 |
|
|
Shashank Shekhar
|
3ccbe08626
|
Fixed reward_clipping and added first test
|
2018-09-03 23:31:12 +05:30 |
|
|
Shashank Shekhar
|
2df0a15a6c
|
added reward clipping wrapper
|
2018-09-01 19:36:00 +05:30 |
|
|
Ryan Curtin
|
aab0b3d81b
|
Fix wrong variable usage.
|
2018-09-01 00:00:58 -04:00 |
|
|
akhandait
|
f5ae73538f
|
Correct beta Backward() function, documentation edit.
|
2018-08-29 20:07:50 +05:30 |
|
|
Ryan Curtin
|
f829451547
|
Merge pull request #1502 from rcurtin/lmnn-test-fix
Use 2 neighbors for LMNNAccuracyTest because 1 is unstable.
|
2018-08-24 23:20:22 -04:00 |
|
|
gmanlan
|
ebe1d035c0
|
+ only publishing artifacts in the build dir
+ moving .exe/.dll to the root installer dir and keeping .lib/.exp in /lib
+ adding dependencies dlls into the installer
|
2018-08-24 17:52:38 -07:00 |
|
|
gmanlan
|
eceb293dcc
|
adding boost and armadillo licenses for the dist.
|
2018-08-24 17:32:35 -07:00 |
|
|
Ryan Curtin
|
6933508df4
|
Use 2 neighbors because 1 is unstable.
It can cause unexpected changes in the impostor, which inverts the gradient. If
we use two impostors and neighbors it is much more robust.
|
2018-08-24 12:24:12 -04:00 |
|
|
Wenhao-H
|
25b8b1ac19
|
comment out RecommendationAccuracySVDPPTest
|
2018-08-24 10:44:08 -04:00 |
|
|
Ryan Curtin
|
22704ad8bc
|
Replace linear regression implementation with something faster.
|
2018-08-23 22:03:33 -04:00 |
|
|
akhandait
|
46224681be
|
Correct MSE Backward() function, add gradient test.
|
2018-08-23 18:47:29 +05:30 |
|
|
akhandait
|
3d5b91f5b9
|
Add simple test.
|
2018-08-23 17:13:56 +05:30 |
|
|
Wenhao-H
|
d590c0b849
|
change default biassvd param
|
2018-08-21 21:24:48 -04:00 |
|
|
Wenhao-H
|
aa06535db3
|
fix gradient computation bug
|
2018-08-21 15:19:06 -04:00 |
|
|
Wenhao-H
|
4f5d182683
|
remove itemRealIdx
|
2018-08-21 15:11:21 -04:00 |
|
|
gmanlan
|
0363841cec
|
pulling doc from jenkins + adding doc & examples into installer
|
2018-08-17 18:07:54 -07:00 |
|
|
gmanlan
|
1f8b498384
|
+version on .msi filename + doxygen install
|
2018-08-15 18:50:46 -07:00 |
|
|
Marcus Edel
|
2be49aea4d
|
Merge pull request #1491 from ShikharJ/Conv
Improve Convolution Gradient() Method.
|
2018-08-15 18:58:22 +02:00 |
|
|
akhandait
|
ed3ee74d1b
|
Add Concatenate layer.
|
2018-08-14 21:16:46 +05:30 |
|
|
Wenhao-H
|
b7a2d9de92
|
debug
|
2018-08-11 17:26:53 -04:00 |
|
|
Wenhao-H
|
d7eeab9720
|
add comments for cf model visitors
|
2018-08-11 12:08:32 -04:00 |
|
|
Wenhao-H
|
ae56e01a29
|
remove @tparam Decompostion
|
2018-08-11 11:58:23 -04:00 |
|
|
Shikhar Jaiswal
|
2c98637391
|
Improve Gradient Method
|
2018-08-11 16:32:25 +05:30 |
|
|
Wenhao-H
|
1d0836dafc
|
fix style and other issues in svd++
|
2018-08-10 07:26:56 +08:00 |
|