Layers

The base layer

This is the class from which all layers inherit.

interface ILayer {
    error TensorTypeNotSupported();
    error IncorrectTensorType();
    error IncorrectTensorDim();

    function appendWeights(Float32x32[] calldata weights, uint idx) external returns (uint, bool);
    function getParamsCount() external view returns (uint);
    function predict(Tensors.TensorData[] calldata input) external returns (Tensors.TensorData memory);    
}

Layer activations

✅ relu ✅ sigmoid ✅ softmax ✅ leakyrelu ✅ tanh ✅ linear

Coming soon:

• softplus

• softsign

Core layers

✅ Input layer ✅ Dense layer ✅ Embedding layer

Coming soon:

• Masking layer

• Lambda layer

• Identity layer

Convolutional Layers

✅ Conv2D layer

✅ Coming soon:

• Conv3D layer

Pooling layers

✅ MaxPooling2D layer ✅ AveragePooling2D layer

Recurrent layers

✅ SimpleRNN layer ✅ LSTM layer

Coming soon:

• GRU layer

• Bidirectional layer

Normalization layers

Coming soon:

• BatchNormalization layer

• LayerNormalization layer

• UnitNormalization layer

• GroupNormalization layer

Reshaping layers

✅ Flatten layer ✅ Rescale layer

Coming soon:

• ZeroPadding2D layer

• Reshape layer

• Permute layer

Merging layers

✅ Add layer ✅ Subtract layer ✅ Multiply layer

Coming soon:

• Concatenate layer

• Average layer

• Maximum layer

• Minimum layer

• Dot layer

Activation Layers

✅ ReLU layer ✅ Sigmoid layer

Coming soon:

• Softmax layer

• LeakyReLU layer

• PReLU layer

• ELU layer

Special Purpose Layers

✅ OnesLike layer ✅ ZerosLike layer