Comments (6)
I didn't know keras has change the API of constraint.
change those lines to the following may be fix the issue.
def get_config(self):
return { "min_value": self.min_value,
"max_value": self.max_value}
from nn_playground.
The repo is updated w.r.t the newest keras.
This line will correctly load saved model.
from nn_playground.
The code here is mostly for demo.
For serious experiments, I got the numbers by strictly following the paper (try my best) :
datasets | paper reported | squared hinge loss | crossentropy loss |
---|---|---|---|
MNIST (MLP) | 0.96% | 1.06% | 1.02% |
CIFAR10 (CNN) | 11.40% | 11.92% | 11.91% |
SVHN (CNN) | 2.80% | 2.94% | 2.82% |
from nn_playground.
Thanks. That worked! :)
from nn_playground.
One more thing. Did you compare the accuracy of your network with the one described in the paper? The paper says they were able to reach a test error of 0.96%.
I trained your network with 3 layers, 1000 epochs and 4096 units and the following result
epoch | acc | loss | val_acc | val_loss
999 | 0.987000008523 | 0.0106547690936 | 0.969300017953 | 0.200085629746
Why do you think the error rate is higher? Also it took 20 hours for 1000 epochs on a Tesla K80 GPU.
from nn_playground.
How are you calculating the test error percentage?
from nn_playground.
Related Issues (17)
- GCNN HOT 5
- Unused code block in qrnn.py HOT 1
- How are the results? Do you obtain the same results?
- qrnn error=> TypeError: unsupported operand type(s) for +: 'NoneType' and 'int' HOT 1
- use_bias = True Error in binary_layers.py
- ternary_tanh
- BinaryConv1D
- Have you see the different when use of normalizing method as pre-processing or not.
- Error in running imdb_lm_gcnn.py HOT 3
- BNN taking longer time than Full Precision
- BNN saved model weights HOT 4
- XNOR-Net output size problem
- binarynet weights are not binary after trained HOT 3
- update to newest version
- ValueError: None values not supported. HOT 1
- states_value parameter is missing from QRNN HOT 2
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from nn_playground.