Similar to pytorch w/ sequential network composed of layers (w/ separate activations) and trainers using strategy pattern instead of being built into the the NN
This would allow for LinearLayer (currently Layer) and in the future Convolution/MaxPool/Attention layers
Similar to pytorch w/ sequential network composed of layers (w/ separate activations) and trainers using strategy pattern instead of being built into the the NN
This would allow for LinearLayer (currently
Layer) and in the future Convolution/MaxPool/Attention layers