Knowledge Editing for LLMs Papers
A must-read paper list on knowledge editing for large language models, tracking surveys, benchmarks, and techniques for updating a trained model's knowledge without full retraining.
Knowledge Editing for LLMs Papers is a running collection of must-read papers on knowledge editing for large language models, the problem of updating, correcting, or removing specific facts a trained model holds without retraining it from scratch. Entries include surveys and benchmarks, techniques such as the KnowEdit benchmark and the EasyEdit framework, and applications such as detoxifying a model or editing knowledge-graph embeddings that a model encodes.
The list also tracks conference tutorials and invited talks on the topic, such as sessions at AAAI, AACL, COLING, IJCAI, CCL, and VALSE, each linked to its slide deck, alongside papers accepted at venues including NeurIPS, EMNLP, and ACL.
It is maintained as a reading list rather than runnable software, and is MIT licensed.