Movement Pruning is a simple, deterministic first-order weight pruning method that is more adaptive to pretrained model fine-tuning. Magnitude pruning can be seen as utilizing zeroth-order information (absolute value) of the running model. In contrast, movement pruning methods are where importance is derived from first-order information. Intuitively, instead of selecting weights that are far from zero, we retain connections that are moving away from zero during the training process.
Source: Movement Pruning: Adaptive Sparsity by Fine-TuningPaper | Code | Results | Date | Stars |
---|
Task | Papers | Share |
---|---|---|
Network Pruning | 2 | 33.33% |
Model Compression | 1 | 16.67% |
Text Generation | 1 | 16.67% |
Machine Translation | 1 | 16.67% |
Question Answering | 1 | 16.67% |
Component | Type |
|
---|---|---|
🤖 No Components Found | You can add them if they exist; e.g. Mask R-CNN uses RoIAlign |