SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2310.20150 · EMNLP · 2023

Unlearn What You Want to Forget: Efficient Unlearning for LLMs

Jiaao Chen, Diyi Yang

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 7 functions out of this paper's own repositories and ran 2 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

RepositoryRoleRan
SALT-NLP/Efficient_Unlearning canonical 2 of 7
FunctionStatusWhere it lives
get_subnet_constructor Ran SALT-NLP/Efficient_Unlearning/src/models/transformers/parameter-efficient-finetuning/modeling.py
code served (permissive licence) · get_code("4eda9cf07aec0693")
kronecker_product Ran SALT-NLP/Efficient_Unlearning/src/models/transformers/parameter-efficient-finetuning/modeling.py
code served (permissive licence) · get_code("54fa343f001c7cff")
adjust_tensors_for_parallel Not yet run SALT-NLP/Efficient_Unlearning/src/models/transformers/parameter-efficient-finetuning/composition.py
code served (permissive licence) · get_code("f8ef77acae8f9f42")
build_full_config Not yet run SALT-NLP/Efficient_Unlearning/src/models/transformers/parameter-efficient-finetuning/configuration.py
code served (permissive licence) · get_code("e68e49764b72c7c1")
get_head_config_and_rename_list Not yet run SALT-NLP/Efficient_Unlearning/src/models/transformers/parameter-efficient-finetuning/head_utils.py
code served (permissive licence) · get_code("2e2d39504f7b42e7")
parse_composition Not yet run SALT-NLP/Efficient_Unlearning/src/models/transformers/parameter-efficient-finetuning/composition.py
code served (permissive licence) · get_code("75cb837427e08cea")
parse_heads_from_composition Not yet run SALT-NLP/Efficient_Unlearning/src/models/transformers/parameter-efficient-finetuning/composition.py
code served (permissive licence) · get_code("302b48be3a814ec4")

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Large language models (LLMs) have achieved significant progress from pre-training on and memorizing a wide range of textual data, however, this process might suffer from privacy issues and violations of data protection regulations. As a result, the ability to easily remove data related to individual users from such models while not deteriorating their predictive quality after the removal becomes increasingly important. To address these issues, in this work, we propose an efficient unlearning framework that could efficiently update LLMs without having to retrain the whole model after data removals, by introducing lightweight unlearning layers learned with a selective teacher-student objective into the transformers. In addition, we introduce a fusion mechanism to effectively combine different unlearning layers that learns to forget different sets of data to handle a sequence of forgetting operations. Experiments on classification and generation tasks demonstrate the effectiveness of our proposed methods compared to the state-of-the-art baselines 1 .

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2310.20150")
get_code_for_paper("2310.20150")
have("2310.20150")

Connect an agent — have() is free.