Recurrent neural network → sequences of information, when learning sequences can start producing own information
Applications of AI? what is linear regression
Start with linear and activation function
Activation function -> nonlinear functions Convolution -> average Convolutions -> averaging into n dimensions Relu Universal approximation theorem adam@deepgram.com blog.deepgram.com Loss function: how well did we do? Mean-squared, cross-entropy, CTC, choice of loss function usually affects last model layer. Optimizer how do we get batter? SGD, RMSProp, Adam Softmax denselayer
Kernel in terms of nvidia Kur Parameters are trained hyper are use to train models
‘virtualenv -p python3 env’ source bin/activate Install ~/.zshrc bash line -> just in general.
Zsh_theme = bullet-train Oh my zen shell Pypy vs pip -> pip asks pypy pip freeze Yenn lecun
Strides in deep learning