Pruning can be done using various algorithms, including:
- L1 norm pruning: Removing weights with a low L1 norm (i.e., weights with a small absolute value).
- L2 norm pruning: Removing weights with a low L2 norm (i.e., weights with a small magnitude).
- Relu pruning: Removing neurons with a low ReLU (i.e., neurons with a small output value).
- Gelu pruning: Removing neurons with a low Gelu (i.e., neurons with a small output value).