Davis King
060e876b86
Made the mex wrapper flush output streams to the matlab console when the mex
...
function ends.
2016-04-20 08:39:49 -04:00
Davis King
168574bd6c
Added visit_layer_parameter_gradients() and also fixed a silly synchronization
...
error in the multi-gpu training code.
2016-04-19 06:45:35 -04:00
Davis King
d31723ff45
Fixed typo in example
2016-04-19 06:44:31 -04:00
Davis King
b73f5e8160
merged
2016-04-18 22:49:04 -04:00
Davis King
b16cc99e8f
Added comments about using multiple GPUs
2016-04-18 22:48:07 -04:00
Davis King
ca11d10806
Added multi-gpu support to the dnn_trainer
2016-04-18 22:03:02 -04:00
Davis King
b9fd956487
Added missing assert
2016-04-18 22:02:42 -04:00
Davis King
172577d2b8
Added find_optimal_parameters()
2016-04-18 18:11:38 -04:00
Davis King
8a7f6873be
Fixed typo in error message
2016-04-18 18:11:16 -04:00
Davis King
8984f75469
Added operator= that takes an initializer_list so that = assignments always
...
behave in the way you would expect.
2016-04-18 18:04:00 -04:00
Davis King
e9fa65398f
Added a matrix constructor that takes an initializer list.
2016-04-18 17:35:02 -04:00
Davis King
3b35f55baf
Added visit_layer_parameters()
2016-04-17 14:39:41 -04:00
Davis King
1fc117e1ab
Clarified error message
2016-04-17 14:35:40 -04:00
Davis King
d85de930a4
Split the update() methods into two parts. One that computes gradients
...
with respect to parameters and one that updates the parameters with those
gradients.
2016-04-17 10:11:05 -04:00
Davis King
8c64a656a2
fixed linker error
2016-04-17 07:41:29 -04:00
Davis King
e1367f1b0f
Made unit test less likely to false alarm.
2016-04-16 22:43:17 -04:00
Davis King
2703a72dfe
Added get_num_devices()
2016-04-16 21:12:31 -04:00
Davis King
603d474352
- Renamed network_type::num_layers to network_type::num_computational_layers.
...
- Made layer() recurse into repeat objects so that the index given to layer()
does what you would expect.
- Added an operator<< for network objects that prints the network architecture.
2016-04-16 10:50:15 -04:00
Davis King
088546f3f4
fixed typo
2016-04-16 06:36:59 -04:00
Davis King
d25454932c
renamed EXAMPLE_LAYER_ to EXAMPLE_COMPUTATIONAL_LAYER_
2016-04-16 06:36:07 -04:00
Davis King
79adbae2f5
Added a few more tests
2016-04-16 06:22:43 -04:00
Davis King
c48d0973c7
fixed grammar in some comments
2016-04-14 21:48:59 -04:00
Davis King
13bc3880e1
Fixed compiler errors when using std::vector in mex wrapper.
2016-04-12 09:20:09 -04:00
Davis King
61591b13e2
Seeded random number generator with the clock since that's generally a good
...
thing to do for this kind of training.
2016-04-11 23:11:18 -04:00
Davis King
02c27ff916
fixed formatting
2016-04-11 23:06:32 -04:00
Davis King
2e39e5fa22
Clarified spec and cleaned up code slightly
2016-04-11 22:58:54 -04:00
Davis King
423cd85594
renamed a file
...
--HG--
rename : examples/dnn_mnist_resnet_ex.cpp => examples/dnn_mnist_advanced_ex.cpp
2016-04-11 22:57:11 -04:00
Davis King
902a2beeaf
Fleshed out these examples more.
2016-04-11 22:55:49 -04:00
Davis King
02b844ea5c
Fixed grammar and clarified a few things.
2016-04-11 21:18:14 -04:00
Davis King
d6c603096d
fixed an array indexing error.
2016-04-11 09:00:44 -04:00
Davis E. King
c3a9a7c463
Merge pull request #94 from e-fominov/fhog-simd8f
...
Added simd8 usage on FHOG features computation
2016-04-11 06:37:17 -04:00
Fm
446a06a035
Removed simd4 and ifdef's
2016-04-11 11:08:52 +03:00
Davis King
7d7c932f29
Added a narrative to this example.
2016-04-10 17:30:45 -04:00
Davis King
6dbc78df03
Fixed errors in documentation
2016-04-10 17:30:07 -04:00
Davis King
f75e2dbf7a
Minor code cleanup
2016-04-10 13:41:05 -04:00
Davis King
67a81c1c51
Made examples work with new fc<> template.
2016-04-10 12:11:19 -04:00
Davis King
feb81e1f4f
Made fc<> always use a bias and added an fc_no_bias<> for the other case.
2016-04-10 12:10:56 -04:00
Davis King
6acddf99de
Just renamed variables to reflect the new meaning of the batch normalization
...
running variance output.
2016-04-10 11:31:24 -04:00
Davis King
538de238fa
merged
2016-04-10 10:54:49 -04:00
Davis King
f9cb3150d0
upgraded to cudnn v5. Also changed the affine_ layer to not be templated but
...
to automatically select the right mode. The serialization format for bn_
layers has also changed, but the code will still be able to deserialize older
bn_ objects.
2016-04-10 10:52:40 -04:00
Davis King
5dad959a02
Made the cuda test include some header files to make it a little more
...
realistic.
2016-04-09 23:10:27 -04:00
Davis King
c81825c3c2
Made unit test more robust
2016-04-09 19:54:59 -04:00
Davis King
9cd103e545
Minor changes to avoid warnings and a bug in clang.
2016-04-09 12:53:12 -04:00
Davis King
b509a16943
Made tests more robust.
2016-04-09 10:07:33 -04:00
Davis King
5fecb30c99
Fixed c++11 test program.
2016-04-09 09:21:17 -04:00
Davis King
945c62bf93
Adjusted test threshold to avoid false test failure.
2016-04-09 09:17:55 -04:00
Davis King
b47ab37b24
Improved C++11 detection.
2016-04-09 09:16:57 -04:00
Davis King
6e9ee89ef6
merged
2016-04-08 23:13:48 -04:00
Davis King
fe168596a2
Moved most of the layer parameters from runtime variables set in constructors
...
to template arguments. This way, the type of a network specifies the entire
network architecture and most of the time the user doesn't even need to do
anything with layer constructors.
2016-04-08 23:12:53 -04:00
Davis King
001bca78e3
Improved the add_layer forwarding constructor. Also added repeat_group()
...
variable decorator.
2016-04-08 22:37:28 -04:00
