59
GitHub - siekmanj/sieknet: A neural network/deep learning library implemented fr...
source link: https://github.com/siekmanj/sieknet
Go to the source link to view the article. You can view the picture content, updated content and better typesetting reading experience. If the link is broken, please click the button below to view the snapshot at that time.
README.md
sieknet
This is a dependency-free neural network/deep learning library I am attempting to write from scratch in C.
I have used the following resources extensively in the course of this project:
- Michael Nielsen's excellent online book
- 3blue1brown's YouTube series
- Wikipedia's article on Recurrent Neural Networks
- Andrej Karpathy's article on RNN's
- Eli Bendersky's article on the softmax function
So far, I have implemented:
- A feed-forward neural network which you can find under
/networks/MLP/
. - A recurrent neural network, which you can find under
/networks/RNN/
. - Backpropagation.
- Softmax activation and cross-entropy cost.
Various demonstrations of how to use the networks can be found in /examples/
. Currently, I am using Shakespeare's sonnets to train a 512x512x512 recurrent network to compose a sonnet.
To do a quick-and-dirty demo:
git clone https://github.com/siekmanj/sieknet
cd sieknet/build
make recurrent_demo && ./recurrent_demo.out
In the coming months, I intend to implement:
- A LSTM network.
- A convolutional neural network.
- Geoffrey Hinton's Capsule network
- Filip Pieknewski's Predictive Vision Model
Recommend
About Joyk
Aggregate valuable and interesting links.
Joyk means Joy of geeK