Skip to content

Latest commit

 

History

2 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

LEAP

Learning Edit-Adaptable Parameters for Multi-Hop Knowledge Propagation

LEAP is a MAML-based meta-training framework that learns edit-adaptable LoRA parameters so that ordinary single-hop knowledge edits also propagate to their multi-hop consequences, without changing the test-time editing procedure itself.

Structure

  • meta-training/ — meta-trains the edit-adaptable LoRA adapter (meta_adapter).
  • evaluation/ — evaluates a meta_adapter on MQuAKE multi-hop edit propagation.
  • analysis/ — interpretability analysis (logit lens, activation patching) of how LEAP changes edit propagation internally.

License

Code is released under the MIT License (see LICENSE). Parts of evaluation/ and analysis/ are adapted from CaKE (MIT License, Copyright (c) 2025 ZJUNLP) — see THIRD_PARTY_LICENSES/ and each subdirectory's README for details.

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages