Layer
Layer in Deep learning can be either a mathematical formula or a process. mathematical formula layers either have weights such as Dense Layer, Conv2D Layer, LSTM Layer or don't have weights such as Activation Layers.
Some layers that are not mathematical formulas are Concatenation Layer, Flatten Layer, Reshape Layer.
Weights
Weights in Deep Learning are numbers that are multiplied by input, and these weights are updated during Backpropagation, using gradient descend.
Note: Some people only consider layers with weights as layers.
