Usually in neural networks, each neuron has their own weights with different values that will be tuned during training.

Shared weight here means that every neuron in the network shares the exact same weight value.