 DricomDragon
|
95f45be5db
Add argument parser for learning rate and epochs
|
5 years ago |
|
DricomDragon
|
b5ac958003
Implement minibatch shuffle
|
5 years ago |
|
DricomDragon
|
b18be6b40b
Use adapted gaussian deviation to initialize weights
|
5 years ago |
|
DricomDragon
|
e8cc1c94ba
Create every weight generator
|
5 years ago |
|
DricomDragon
|
eedb7117ff
Make weight generator more user-friendly
|
5 years ago |
|
DricomDragon
|
d9f1f7425b
Use gaussian distribution to initialize weights
|
5 years ago |
|
DricomDragon
|
c1a9d30e6c
Fix bug due to implicit copy
|
5 years ago |
|
DricomDragon
|
91e01907ea
Fix misprint in ratio
|
5 years ago |
|
DricomDragon
|
8f4edddc63
Fix dimension issue
|
5 years ago |
|
DricomDragon
|
3a1319e534
Reduce number of epochs for testing purpose
|
5 years ago |
|
DricomDragon
|
032de4ba78
Fix execution of user script
|
5 years ago |
|
DricomDragon
|
95195658a1
Implement testing sample loading
|
5 years ago |
|
DricomDragon
|
8e9fc3c688
Implement benchmark
|
5 years ago |
|
DricomDragon
|
1d5db18dc3
Implement training
|
5 years ago |
|
DricomDragon
|
8064c8acc6
Transpose outputs of sample loading
|
5 years ago |
|
DricomDragon
|
5f7d7872f6
Create neural network structure
|
5 years ago |
|
DricomDragon
|
172a3723aa
Change returned label value as numpy array
|
5 years ago |
|
DricomDragon
|
9bc1aa4926
Write the high level structure of source code
|
5 years ago |
|
DricomDragon
|
6411e2564c
Add computing python dependencies
|
5 years ago |
|
DricomDragon
|
70d36c5af5
Merge branch 'mnist'
|
5 years ago |
|
DricomDragon
|
6d587b4c38
Create function to load training data samples
|
5 years ago |
|
DricomDragon
|
2ae3b16636
Describe requirements to use source code
|
5 years ago |
|
DricomDragon
|
8f835857c0
Import example code from README
|
5 years ago |
|
DricomDragon
|
58168e6b0d
Create a script to retrieve digit database
|
5 years ago |
 Jovian Hersemeule
|
236aca57a1
Initial commit
|
5 years ago |
