{"about":{"non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","site":"https://codewithpapers.app","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page","syntology":{"site":"https://syntology.ai","developers":"https://syntology.ai/developers","mcp":{"server":"https://syntology.ai/mcp","transport":"streamable-http","server_card":"https://syntology.ai/.well-known/mcp/server-card.json","auth":{"type":"trial token, no account","trial_token":"https://syntology.ai/api/oauth/trial/token","method":"POST","docs":"https://syntology.ai/developers"}},"have":"https://syntology.ai/api/graph/have?x=<method, arXiv id or title> (free, answers coverage only)","paper_base":"https://syntology.ai/paper/","atlas_base":"https://app.syntology.ai/?focus="},"machine_readable":[{"url":"https://codewithpapers.app/llms.txt","what":"the machine catalog: every machine-readable file, counted"},{"url":"https://codewithpapers.app/index/manifest.json","what":"paper-to-code index by arXiv id, with Syntology's counts"},{"url":"https://codewithpapers.app/search/manifest.json","what":"site search index (titles, authors) and its files"},{"url":"https://codewithpapers.app/download","what":"bulk files: Syntology's layer, described there"},{"url":"https://codewithpapers.app/build_manifest.json","what":"the build record: inputs, counts, exclusions, probes"}]},"url":"/paper/knowledge-unlearning-for-mitigating-privacy","title":"Knowledge Unlearning for Mitigating Privacy Risks in Language Models","arxiv_id":"2210.01504","date":"2022-10-04","proceeding":null,"authors":["Joel Jang","Dongkeun Yoon","Sohee Yang","Sungmin Cha","Moontae Lee","Lajanugen Logeswaran","Minjoon Seo"],"abstract":"Pretrained Language Models (LMs) memorize a vast amount of knowledge during initial pretraining, including information that may violate the privacy of personal lives and identities. Previous work addressing privacy issues for language models has mostly focused on data preprocessing and differential privacy methods, both requiring re-training the underlying LM. We propose knowledge unlearning as an alternative method to reduce privacy risks for LMs post hoc. We show that simply performing gradient ascent on target token sequences is effective at forgetting them with little to no degradation of general language modeling performances for larger LMs; it sometimes even substantially improves the underlying LM with just a few iterations. We also find that sequential unlearning is better than trying to unlearn all the data at once and that unlearning is highly dependent on which kind of data (domain) is forgotten. By showing comparisons with a previous data preprocessing method and a decoding method known to mitigate privacy risks for LMs, we show that unlearning can give a stronger empirical privacy guarantee in scenarios where the data vulnerable to extraction attacks are known a priori while being much more efficient and robust. We release the code and dataset needed to replicate our results at https://github.com/joeljang/knowledge-unlearning.","url_abs":"https://arxiv.org/abs/2210.01504v2","url_pdf":"https://arxiv.org/pdf/2210.01504v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"knowledge-unlearning-for-mitigating-privacy","repo_url":"https://github.com/joeljang/knowledge-unlearning","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"knowledge-unlearning-for-mitigating-privacy","repo_url":"https://github.com/shreya1313/llm-unlearning","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"language-modeling","task_name":"Language Modeling"},{"task_slug":"language-modelling","task_name":"Language Modelling"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/language-modelling-on-the-pile","task":"Language Modelling","dataset":"The Pile","model":"GPT-Neo 2.7B","rank_in_archive_order":30,"of":39,"metrics":{"Test perplexity":"10.44"},"uses_additional_data":false},{"leaderboard":"/sota/language-modelling-on-the-pile","task":"Language Modelling","dataset":"The Pile","model":"GPT-Neo 1.3B","rank_in_archive_order":32,"of":39,"metrics":{"Test perplexity":"11.46"},"uses_additional_data":false},{"leaderboard":"/sota/language-modelling-on-the-pile","task":"Language Modelling","dataset":"The Pile","model":"OPT 2.7B","rank_in_archive_order":34,"of":39,"metrics":{"Test perplexity":"17.81"},"uses_additional_data":false},{"leaderboard":"/sota/language-modelling-on-the-pile","task":"Language Modelling","dataset":"The Pile","model":"GPT-Neo 125M","rank_in_archive_order":35,"of":39,"metrics":{"Test perplexity":"17.83"},"uses_additional_data":false},{"leaderboard":"/sota/language-modelling-on-the-pile","task":"Language Modelling","dataset":"The Pile","model":"OPT 1.3B","rank_in_archive_order":36,"of":39,"metrics":{"Test perplexity":"19.55"},"uses_additional_data":false},{"leaderboard":"/sota/language-modelling-on-the-pile","task":"Language Modelling","dataset":"The Pile","model":"OPT 125M","rank_in_archive_order":38,"of":39,"metrics":{"Test perplexity":"32.26"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/2210.01504","atlas_url":"https://app.syntology.ai/?focus=2210.01504","mcp":null,"developers":"https://syntology.ai/developers","held":true,"reason":"arxiv_id_in_atlas","checked_against":{"atlas_ids":549449,"atlas_ids_sha256":"afd72399dafd9906e9062927f70655952385a3d75488558e1f0484e950ab09b7","read_at":"2026-10-05T01:46:21Z"},"agent_calls":[{"tool":"get_citation_path","arguments":{"paper_1":"2210.01504"},"arguments_in":{"paper_2":"another paper's arXiv id or title"},"call":"get_citation_path(paper_1=\"2210.01504\", paper_2=\"…\")"},{"tool":"get_concepts_for_paper","arguments":{"arxiv_id":"2210.01504"},"call":"get_concepts_for_paper(arxiv_id=\"2210.01504\")"}],"by_repo":[{"repo":"joeljang/knowledge-unlearning","kind":"official","samples":0,"ran":0,"instrument":0,"unverified":0,"harvested_for_other_papers":1,"harvested_for":["2601.22359"],"harvested_for_calls":[{"arxiv_id":"2601.22359","tool":"get_harvested_code_for_paper","arguments":{"arxiv_id":"2601.22359"},"call":"get_harvested_code_for_paper(arxiv_id=\"2601.22359\")"}]},{"repo":"shreya1313/llm-unlearning","kind":"listed","samples":0,"ran":0,"instrument":0,"unverified":0}],"by_repo_harvested_for_is":"harvested_for: the arXiv ids (as Syntology's graph holds them) of every paper whose samples Syntology's graph links to this repository, in arXiv-id order, complete; the page names the first 3 and says how many more. A link only: it does not say the papers are the same work, that one copied the other, or that the code implements this paper's method; harvested_for_calls gives the get_harvested_code_for_paper call for each arXiv-shaped id among them","by_repo_is":"one row per repository in the archive's code links, Syntology's graph links or the samples; samples 0 means no sample is linked to this paper, a statement about Syntology's coverage, not about the repository: with harvested_for_other_papers present, Syntology harvested the repository for that many other papers, and without it, Syntology harvested nothing from it; placed_by_identical_code counts samples whose link records no repository, placed here because identical code was harvested from this repository for another paper; ran counts executions on a synthesized input and instrument counts failures of Syntology's instrument, not of the code"},"code_links_note":null,"arxiv_metadata":null,"syntology_extracted_results":null,"record_sha256":"528705efa9ad3c71bfbe1d56a24dde358fe9f958828a79b6edd3cd8b5cb1d21d","record_changed_at":"2026-09-28","record_changed_at_basis":"first_hashed"}