colin
|
206ba35c93
|
refine export ,add layer2-4
|
2020-07-28 13:50:58 +08:00 |
|
colin
|
ce1aca9531
|
mv shape to single file
|
2020-07-28 13:35:50 +08:00 |
|
colin
|
e5410cb1b2
|
add shape output and CheckResnet50
|
2020-07-27 20:27:00 +08:00 |
|
colin
|
e2e2b01b54
|
add layer1 inner print
|
2020-07-27 16:30:54 +08:00 |
|
colin
|
6309c2f066
|
fix auto layer1 lose two layer
|
2020-07-24 14:03:07 +08:00 |
|
colin
|
7e3d3bab6f
|
remove unuse print
|
2020-07-23 17:07:18 +08:00 |
|
colin
|
7a3dfde615
|
fix resnet to eval static , fix batch normal error
|
2020-07-23 15:26:40 +08:00 |
|
colin
|
ee0a12f7dd
|
add verify resnet
|
2020-07-20 17:35:05 +08:00 |
|
colin
|
039310704e
|
fix byte count with float * 4
|
2020-07-20 16:00:32 +08:00 |
|
colin
|
3d062150e9
|
add input and output in resnet50 demo
|
2020-07-20 15:45:02 +08:00 |
|
colin
|
a14983b756
|
add bn weight and bias
|
2020-07-20 13:05:12 +08:00 |
|
colin
|
7ae864fe48
|
add ; in cc
|
2020-07-20 11:21:02 +08:00 |
|
colin
|
ef0dd5bc00
|
add ResNet weight dump
|
2020-07-18 11:23:58 +08:00 |
|
colin
|
7ab55557cc
|
add MnistDemo
|
2020-07-13 20:14:01 +08:00 |
|
colin
|
67c5cc13a0
|
Fix readme
|
2020-03-08 17:11:54 +08:00 |
|
c
|
d3046d1131
|
fix as 4 kernels in first layer.
|
2020-01-02 15:51:24 +08:00 |
|
c
|
346159068c
|
Speedup to search.
1 modle set to channel output 4 from 8
2 network weights is type of int from -4 to 4
|
2019-12-30 17:24:27 +08:00 |
|
c
|
18126824ee
|
Add more train to test entropy resualt.
|
2019-12-30 16:02:40 +08:00 |
|
c
|
e0de909d62
|
Revert "Refine kernal Random to w*h*16 kinds possiable."
This reverts commit 675cd8567e.
|
2019-12-19 16:07:29 +08:00 |
|
c
|
675cd8567e
|
Refine kernal Random to w*h*16 kinds possiable.
Donot use Cpu any more.
|
2019-12-19 15:22:00 +08:00 |
|
c
|
8404d93880
|
Speed up by multi cpu to handle histogram
|
2019-12-18 19:26:42 +08:00 |
|
c
|
418e79df60
|
Add Directed search 8 weights instead of work with two steps.
|
2019-12-18 15:18:01 +08:00 |
|
c
|
fa7a066b87
|
Test 7x7 image of Mnist.
|
2019-12-11 18:47:51 +08:00 |
|
c
|
a368c4eae5
|
Try deep search ,and test in Minist DataSet
|
2019-12-09 11:31:28 +08:00 |
|
c
|
f105932c26
|
Add readme.
|
2019-11-25 14:20:54 +08:00 |
|
Colin
|
8bb2c5562d
|
Add bn layer to 3335,Get if it is benify that.
|
2019-11-06 15:17:14 +08:00 |
colin
|
c2e149fb68
|
Check if train kernel is benifit to pretrain.
|
2019-10-31 14:35:47 +08:00 |
|
colin
|
8a9acaa94e
|
New test
|
2019-10-15 15:26:03 +08:00 |
|
colin
|
febad1f9c3
|
Fix filter must sum=0
|
2019-10-03 19:50:45 +08:00 |
|
colin
|
9628b2c17e
|
add minactive return in search
|
2019-10-03 16:30:52 +08:00 |
|
colin
|
130a6942b9
|
refine Search Weight calculater time.
|
2019-09-30 18:33:47 +08:00 |
|
colin
|
76c9b7f05e
|
Refine line display visdom
|
2019-09-23 10:02:01 +08:00 |
|
colin
|
7f39f884fc
|
Refine Distance of kernels
|
2019-09-20 12:48:45 +08:00 |
|
colin
|
b0246d2a95
|
Refine find best 8
|
2019-09-17 23:20:24 +08:00 |
|
colin
|
f858ae1821
|
Add UnsuperLearnFindBestWeight to find best from 4000
|
2019-09-14 21:02:56 +08:00 |
|
colin
|
6201163e14
|
Add UnsuperLearnTrainWeight to train new weight kernel
|
2019-09-02 19:28:04 +08:00 |
|
colin
|
d4a22775d4
|
Add two method of real wit kernel genarator.
1 get 8 from 4000 with entropy calculation.
2 Auto Grad to regrethion kernel.
|
2019-09-01 20:23:00 +08:00 |
|
colin
|
5220bf6885
|
start to creat 8 from 4000
|
2019-08-27 23:06:24 +08:00 |
|
colin
|
d9df29f04a
|
train with pretrain
|
2019-08-23 23:27:45 +08:00 |
|
colin
|
81d8931842
|
init git
|
2019-08-19 15:53:10 +08:00 |