LwPTV prunes low-saliency layers of task vectors before merging, improving OOD accuracy of merged models while roughly maintaining ID accuracy.
When is task vector provably effective for model editing? a generalization analysis of nonlinear transformers
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
Merging Smarter, Generalizing Better: Enhancing Model Merging on OOD Data
LwPTV prunes low-saliency layers of task vectors before merging, improving OOD accuracy of merged models while roughly maintaining ID accuracy.