← Back to Regularization

structured pruning

Regularization
Used in
4 PRs
Best BPB
1.0850
Avg BPB
1.1262

Hyperparameters Across PRs

pr_numberparameters
1019{"type":"±1 by reconstruction error"}
1551{"top_k":true,"gradual_budget_annealing":true}
1849{"target":"MLP hidden channels","strategy":"per-block capped pruning","ablation":"zero selected fc rows and matching proj columns"}
1849{"target":"MLP hidden channels","strategy":"soft-cap pruning","score_weights":{"activation_weighted_score":0.7,"norm_score":0.3},"local_rank_weight":0.75,"cap_multiplier":1.75,"floor_multiplier":0}