详细文档请参照:https://blog.csdn.net/kid_14_12/article/details/105852626
待添加.....
封装numpy数组,是NumpyFlow的数据载体,相当于torch中的Tensor。
- Operation:用于支持基本的运算及对应的梯度计算,是支持自动微分的基本算子
- Tensor:
- Optimizer
- Module
- Assign、Add、Multiply、Subtract
- Divide、Negative、Positive、Power
- Exp、Log、Log2、Log10
- MatMul、EinSum
- sgd
- adam
- Kaiming初始化
- Xavier初始化
- 随机初始化
- 填充0初始化
- 填充1初始化
- Linear
- Relu、Sigmid、Softmax
- Conv1D
- Conv2D
- Conv3D
- MaxPool1D
- MaxPool2D
- MaxPool3D
- BatchNorm1D
- BatchNorm2D
- BatchNorm3D
- Reshape
- Permute
- Flatten
- RepeatVector
- Lambda
- ActivityRegularization
- Masking
- SpatialDropout1D
- SpatialDropout2D
- SpatialDropout3D
- LocallyConnected1D
- LocallyConnected2D
- RNN
- GRU
- LSTM
- ConvLSTM2D
- SimpleRNNCell
- GRUCell
- LSTMCell
- CuDNNGRU
- CuDNNLSTM
- Embedding