This repository has been archived by the owner on Aug 13, 2022. It is now read-only.
io.github.alexhaoge.mydl.mydl 1.0-alpha Latest version
Install 1/2: Add this to pom.xml:
Learn more about Maven or Gradle
<dependency>
<groupId>io.github.alexhaoge.mydl</groupId>
<artifactId>mydl</artifactId>
<version>1.0-alpha</version>
</dependency>
Install 2/2: Run via command line
$ mvn install
About this package
This is a pre-release of MyDL, a lightweight deep learning framework entirely based on Java.
Because the project is still under construction. This release may be unstable.
If you meet any problems, please feel free to raise an issue or contact us and we will fix it as soon as possible.
- Model: abstract model, sequential
- Layer: abstract layer
- Activation: abstract activation, ReLU, Sigmoid, Tanh, Softmax
- Fully-connected: Dense(for high dimension), Linear1D
- Reshape layer
- Loss: abstract loss, SSE, MSE, binary cross-entropy, categorical cross-entropy
- Optimizer: abstract optimizer, mini-batch GD/SGD
- Tensor: abstract Tensor, Tensor1D, Tensor2D, Tensor3D
- Tensor_size(record the shape of tensor)
- Dataset: APIs to load MNIST dataset(included in JAR)
- Utils:
- Data(self-defined class for model training, easy to convert)
- IsAttribute(Determin if the object has such attribute)
