Learning to Edit: Aligning LLMs with Knowledge Editing Article Swipe
YOU?
·
· 2024
· Open Access
·
· DOI: https://doi.org/10.48550/arxiv.2402.11905
Knowledge editing techniques, aiming to efficiently modify a minor proportion of knowledge in large language models (LLMs) without negatively impacting performance across other inputs, have garnered widespread attention. However, existing methods predominantly rely on memorizing the updated knowledge, impeding LLMs from effectively combining the new knowledge with their inherent knowledge when answering questions. To this end, we propose a Learning to Edit (LTE) framework, focusing on teaching LLMs to apply updated knowledge into input questions, inspired by the philosophy of "Teach a man to fish." LTE features a two-phase process: (i) the Alignment Phase, which fine-tunes LLMs on a meticulously curated parallel dataset to make reliable, in-scope edits while preserving out-of-scope information and linguistic proficiency; and (ii) the Inference Phase, which employs a retrieval-based mechanism for real-time and mass knowledge editing. By comparing our approach with seven advanced baselines across four popular knowledge editing benchmarks and two LLM architectures, we demonstrate LTE's superiority in knowledge editing performance, robustness in both batch and sequential editing, minimal interference on general tasks, and rapid editing speeds. The data and code are available at https://github.com/YJiangcm/LTE.
Related Topics
- Type
- preprint
- Language
- en
- Landing Page
- http://arxiv.org/abs/2402.11905
- https://arxiv.org/pdf/2402.11905
- OA Status
- green
- Related Works
- 10
- OpenAlex ID
- https://openalex.org/W4391987803
Raw OpenAlex JSON
- OpenAlex ID
-
https://openalex.org/W4391987803Canonical identifier for this work in OpenAlex
- DOI
-
https://doi.org/10.48550/arxiv.2402.11905Digital Object Identifier
- Title
-
Learning to Edit: Aligning LLMs with Knowledge EditingWork title
- Type
-
preprintOpenAlex work type
- Language
-
enPrimary language
- Publication year
-
2024Year of publication
- Publication date
-
2024-02-19Full publication date if available
- Authors
-
Yuxin Jiang, Yufei Wang, Chuhan Wu, Wanjun Zhong, Xingshan Zeng, Jiahui Gao, Liangyou Li, Xin Jiang, Lifeng Shang, Ruiming Tang, Qun Liu, Wei WangList of authors in order
- Landing page
-
https://arxiv.org/abs/2402.11905Publisher landing page
- PDF URL
-
https://arxiv.org/pdf/2402.11905Direct link to full text PDF
- Open access
-
YesWhether a free full text is available
- OA status
-
greenOpen access status per OpenAlex
- OA URL
-
https://arxiv.org/pdf/2402.11905Direct OA link when available
- Concepts
-
Computer scienceTop concepts (fields/topics) attached by OpenAlex
- Cited by
-
0Total citation count in OpenAlex
- Related works (count)
-
10Other works algorithmically related by OpenAlex
Full payload
| id | https://openalex.org/W4391987803 |
|---|---|
| doi | https://doi.org/10.48550/arxiv.2402.11905 |
| ids.doi | https://doi.org/10.48550/arxiv.2402.11905 |
| ids.openalex | https://openalex.org/W4391987803 |
| fwci | |
| type | preprint |
| title | Learning to Edit: Aligning LLMs with Knowledge Editing |
| biblio.issue | |
| biblio.volume | |
| biblio.last_page | |
| biblio.first_page | |
| topics[0].id | https://openalex.org/T14330 |
| topics[0].field.id | https://openalex.org/fields/17 |
| topics[0].field.display_name | Computer Science |
| topics[0].score | 0.9534000158309937 |
| topics[0].domain.id | https://openalex.org/domains/3 |
| topics[0].domain.display_name | Physical Sciences |
| topics[0].subfield.id | https://openalex.org/subfields/1710 |
| topics[0].subfield.display_name | Information Systems |
| topics[0].display_name | Library Science and Information Systems |
| is_xpac | False |
| apc_list | |
| apc_paid | |
| concepts[0].id | https://openalex.org/C41008148 |
| concepts[0].level | 0 |
| concepts[0].score | 0.4600394368171692 |
| concepts[0].wikidata | https://www.wikidata.org/wiki/Q21198 |
| concepts[0].display_name | Computer science |
| keywords[0].id | https://openalex.org/keywords/computer-science |
| keywords[0].score | 0.4600394368171692 |
| keywords[0].display_name | Computer science |
| language | en |
| locations[0].id | pmh:oai:arXiv.org:2402.11905 |
| locations[0].is_oa | True |
| locations[0].source.id | https://openalex.org/S4306400194 |
| locations[0].source.issn | |
| locations[0].source.type | repository |
| locations[0].source.is_oa | True |
| locations[0].source.issn_l | |
| locations[0].source.is_core | False |
| locations[0].source.is_in_doaj | False |
| locations[0].source.display_name | arXiv (Cornell University) |
| locations[0].source.host_organization | https://openalex.org/I205783295 |
| locations[0].source.host_organization_name | Cornell University |
| locations[0].source.host_organization_lineage | https://openalex.org/I205783295 |
| locations[0].license | |
| locations[0].pdf_url | https://arxiv.org/pdf/2402.11905 |
| locations[0].version | submittedVersion |
| locations[0].raw_type | text |
| locations[0].license_id | |
| locations[0].is_accepted | False |
| locations[0].is_published | False |
| locations[0].raw_source_name | |
| locations[0].landing_page_url | http://arxiv.org/abs/2402.11905 |
| locations[1].id | doi:10.48550/arxiv.2402.11905 |
| locations[1].is_oa | True |
| locations[1].source.id | https://openalex.org/S4306400194 |
| locations[1].source.issn | |
| locations[1].source.type | repository |
| locations[1].source.is_oa | True |
| locations[1].source.issn_l | |
| locations[1].source.is_core | False |
| locations[1].source.is_in_doaj | False |
| locations[1].source.display_name | arXiv (Cornell University) |
| locations[1].source.host_organization | https://openalex.org/I205783295 |
| locations[1].source.host_organization_name | Cornell University |
| locations[1].source.host_organization_lineage | https://openalex.org/I205783295 |
| locations[1].license | |
| locations[1].pdf_url | |
| locations[1].version | |
| locations[1].raw_type | article |
| locations[1].license_id | |
| locations[1].is_accepted | False |
| locations[1].is_published | |
| locations[1].raw_source_name | |
| locations[1].landing_page_url | https://doi.org/10.48550/arxiv.2402.11905 |
| indexed_in | arxiv, datacite |
| authorships[0].author.id | https://openalex.org/A5100756716 |
| authorships[0].author.orcid | https://orcid.org/0000-0002-8388-7262 |
| authorships[0].author.display_name | Yuxin Jiang |
| authorships[0].author_position | first |
| authorships[0].raw_author_name | Jiang, Yuxin |
| authorships[0].is_corresponding | False |
| authorships[1].author.id | https://openalex.org/A5100374831 |
| authorships[1].author.orcid | https://orcid.org/0000-0002-8108-3818 |
| authorships[1].author.display_name | Yufei Wang |
| authorships[1].author_position | middle |
| authorships[1].raw_author_name | Wang, Yufei |
| authorships[1].is_corresponding | False |
| authorships[2].author.id | https://openalex.org/A5001967239 |
| authorships[2].author.orcid | https://orcid.org/0000-0001-5730-8792 |
| authorships[2].author.display_name | Chuhan Wu |
| authorships[2].author_position | middle |
| authorships[2].raw_author_name | Wu, Chuhan |
| authorships[2].is_corresponding | False |
| authorships[3].author.id | https://openalex.org/A5019101763 |
| authorships[3].author.orcid | https://orcid.org/0009-0007-2236-228X |
| authorships[3].author.display_name | Wanjun Zhong |
| authorships[3].author_position | middle |
| authorships[3].raw_author_name | Zhong, Wanjun |
| authorships[3].is_corresponding | False |
| authorships[4].author.id | https://openalex.org/A5044933782 |
| authorships[4].author.orcid | https://orcid.org/0000-0002-0455-5519 |
| authorships[4].author.display_name | Xingshan Zeng |
| authorships[4].author_position | middle |
| authorships[4].raw_author_name | Zeng, Xingshan |
| authorships[4].is_corresponding | False |
| authorships[5].author.id | https://openalex.org/A5067047605 |
| authorships[5].author.orcid | https://orcid.org/0000-0002-4244-174X |
| authorships[5].author.display_name | Jiahui Gao |
| authorships[5].author_position | middle |
| authorships[5].raw_author_name | Gao, Jiahui |
| authorships[5].is_corresponding | False |
| authorships[6].author.id | https://openalex.org/A5038537171 |
| authorships[6].author.orcid | https://orcid.org/0000-0002-0279-003X |
| authorships[6].author.display_name | Liangyou Li |
| authorships[6].author_position | middle |
| authorships[6].raw_author_name | Li, Liangyou |
| authorships[6].is_corresponding | False |
| authorships[7].author.id | https://openalex.org/A5100650977 |
| authorships[7].author.orcid | https://orcid.org/0009-0005-5164-306X |
| authorships[7].author.display_name | Xin Jiang |
| authorships[7].author_position | middle |
| authorships[7].raw_author_name | Jiang, Xin |
| authorships[7].is_corresponding | False |
| authorships[8].author.id | https://openalex.org/A5046228314 |
| authorships[8].author.orcid | |
| authorships[8].author.display_name | Lifeng Shang |
| authorships[8].author_position | middle |
| authorships[8].raw_author_name | Shang, Lifeng |
| authorships[8].is_corresponding | False |
| authorships[9].author.id | https://openalex.org/A5114191001 |
| authorships[9].author.orcid | |
| authorships[9].author.display_name | Ruiming Tang |
| authorships[9].author_position | middle |
| authorships[9].raw_author_name | Tang, Ruiming |
| authorships[9].is_corresponding | False |
| authorships[10].author.id | https://openalex.org/A5100426170 |
| authorships[10].author.orcid | https://orcid.org/0000-0002-7000-1792 |
| authorships[10].author.display_name | Qun Liu |
| authorships[10].author_position | middle |
| authorships[10].raw_author_name | Liu, Qun |
| authorships[10].is_corresponding | False |
| authorships[11].author.id | https://openalex.org/A5100688855 |
| authorships[11].author.orcid | https://orcid.org/0000-0002-6104-7908 |
| authorships[11].author.display_name | Wei Wang |
| authorships[11].author_position | last |
| authorships[11].raw_author_name | Wang, Wei |
| authorships[11].is_corresponding | False |
| has_content.pdf | True |
| has_content.grobid_xml | True |
| is_paratext | False |
| open_access.is_oa | True |
| open_access.oa_url | https://arxiv.org/pdf/2402.11905 |
| open_access.oa_status | green |
| open_access.any_repository_has_fulltext | False |
| created_date | 2025-10-10T00:00:00 |
| display_name | Learning to Edit: Aligning LLMs with Knowledge Editing |
| has_fulltext | False |
| is_retracted | False |
| updated_date | 2025-11-06T06:51:31.235846 |
| primary_topic.id | https://openalex.org/T14330 |
| primary_topic.field.id | https://openalex.org/fields/17 |
| primary_topic.field.display_name | Computer Science |
| primary_topic.score | 0.9534000158309937 |
| primary_topic.domain.id | https://openalex.org/domains/3 |
| primary_topic.domain.display_name | Physical Sciences |
| primary_topic.subfield.id | https://openalex.org/subfields/1710 |
| primary_topic.subfield.display_name | Information Systems |
| primary_topic.display_name | Library Science and Information Systems |
| related_works | https://openalex.org/W2748952813, https://openalex.org/W2390279801, https://openalex.org/W2358668433, https://openalex.org/W2376932109, https://openalex.org/W2001405890, https://openalex.org/W2382290278, https://openalex.org/W2478288626, https://openalex.org/W4391913857, https://openalex.org/W2350741829, https://openalex.org/W2530322880 |
| cited_by_count | 0 |
| locations_count | 2 |
| best_oa_location.id | pmh:oai:arXiv.org:2402.11905 |
| best_oa_location.is_oa | True |
| best_oa_location.source.id | https://openalex.org/S4306400194 |
| best_oa_location.source.issn | |
| best_oa_location.source.type | repository |
| best_oa_location.source.is_oa | True |
| best_oa_location.source.issn_l | |
| best_oa_location.source.is_core | False |
| best_oa_location.source.is_in_doaj | False |
| best_oa_location.source.display_name | arXiv (Cornell University) |
| best_oa_location.source.host_organization | https://openalex.org/I205783295 |
| best_oa_location.source.host_organization_name | Cornell University |
| best_oa_location.source.host_organization_lineage | https://openalex.org/I205783295 |
| best_oa_location.license | |
| best_oa_location.pdf_url | https://arxiv.org/pdf/2402.11905 |
| best_oa_location.version | submittedVersion |
| best_oa_location.raw_type | text |
| best_oa_location.license_id | |
| best_oa_location.is_accepted | False |
| best_oa_location.is_published | False |
| best_oa_location.raw_source_name | |
| best_oa_location.landing_page_url | http://arxiv.org/abs/2402.11905 |
| primary_location.id | pmh:oai:arXiv.org:2402.11905 |
| primary_location.is_oa | True |
| primary_location.source.id | https://openalex.org/S4306400194 |
| primary_location.source.issn | |
| primary_location.source.type | repository |
| primary_location.source.is_oa | True |
| primary_location.source.issn_l | |
| primary_location.source.is_core | False |
| primary_location.source.is_in_doaj | False |
| primary_location.source.display_name | arXiv (Cornell University) |
| primary_location.source.host_organization | https://openalex.org/I205783295 |
| primary_location.source.host_organization_name | Cornell University |
| primary_location.source.host_organization_lineage | https://openalex.org/I205783295 |
| primary_location.license | |
| primary_location.pdf_url | https://arxiv.org/pdf/2402.11905 |
| primary_location.version | submittedVersion |
| primary_location.raw_type | text |
| primary_location.license_id | |
| primary_location.is_accepted | False |
| primary_location.is_published | False |
| primary_location.raw_source_name | |
| primary_location.landing_page_url | http://arxiv.org/abs/2402.11905 |
| publication_date | 2024-02-19 |
| publication_year | 2024 |
| referenced_works_count | 0 |
| abstract_inverted_index.a | 7, 58, 81, 87, 98, 122 |
| abstract_inverted_index.By | 131 |
| abstract_inverted_index.To | 53 |
| abstract_inverted_index.at | 179 |
| abstract_inverted_index.by | 76 |
| abstract_inverted_index.in | 12, 153, 158 |
| abstract_inverted_index.of | 10, 79 |
| abstract_inverted_index.on | 33, 65, 97, 166 |
| abstract_inverted_index.to | 4, 60, 68, 83, 103 |
| abstract_inverted_index.we | 56, 149 |
| abstract_inverted_index.(i) | 90 |
| abstract_inverted_index.LLM | 147 |
| abstract_inverted_index.LTE | 85 |
| abstract_inverted_index.The | 173 |
| abstract_inverted_index.and | 112, 115, 127, 145, 161, 169, 175 |
| abstract_inverted_index.are | 177 |
| abstract_inverted_index.for | 125 |
| abstract_inverted_index.man | 82 |
| abstract_inverted_index.new | 44 |
| abstract_inverted_index.our | 133 |
| abstract_inverted_index.the | 35, 43, 77, 91, 117 |
| abstract_inverted_index.two | 146 |
| abstract_inverted_index.(ii) | 116 |
| abstract_inverted_index.Edit | 61 |
| abstract_inverted_index.LLMs | 39, 67, 96 |
| abstract_inverted_index.both | 159 |
| abstract_inverted_index.code | 176 |
| abstract_inverted_index.data | 174 |
| abstract_inverted_index.end, | 55 |
| abstract_inverted_index.four | 140 |
| abstract_inverted_index.from | 40 |
| abstract_inverted_index.have | 24 |
| abstract_inverted_index.into | 72 |
| abstract_inverted_index.make | 104 |
| abstract_inverted_index.mass | 128 |
| abstract_inverted_index.rely | 32 |
| abstract_inverted_index.this | 54 |
| abstract_inverted_index.when | 50 |
| abstract_inverted_index.with | 46, 135 |
| abstract_inverted_index.(LTE) | 62 |
| abstract_inverted_index.LTE's | 151 |
| abstract_inverted_index.apply | 69 |
| abstract_inverted_index.batch | 160 |
| abstract_inverted_index.edits | 107 |
| abstract_inverted_index.input | 73 |
| abstract_inverted_index.large | 13 |
| abstract_inverted_index.minor | 8 |
| abstract_inverted_index.other | 22 |
| abstract_inverted_index.rapid | 170 |
| abstract_inverted_index.seven | 136 |
| abstract_inverted_index.their | 47 |
| abstract_inverted_index.which | 94, 120 |
| abstract_inverted_index.while | 108 |
| abstract_inverted_index."Teach | 80 |
| abstract_inverted_index.(LLMs) | 16 |
| abstract_inverted_index.Phase, | 93, 119 |
| abstract_inverted_index.across | 21, 139 |
| abstract_inverted_index.aiming | 3 |
| abstract_inverted_index.fish." | 84 |
| abstract_inverted_index.models | 15 |
| abstract_inverted_index.modify | 6 |
| abstract_inverted_index.tasks, | 168 |
| abstract_inverted_index.curated | 100 |
| abstract_inverted_index.dataset | 102 |
| abstract_inverted_index.editing | 1, 143, 155, 171 |
| abstract_inverted_index.employs | 121 |
| abstract_inverted_index.general | 167 |
| abstract_inverted_index.inputs, | 23 |
| abstract_inverted_index.methods | 30 |
| abstract_inverted_index.minimal | 164 |
| abstract_inverted_index.popular | 141 |
| abstract_inverted_index.propose | 57 |
| abstract_inverted_index.speeds. | 172 |
| abstract_inverted_index.updated | 36, 70 |
| abstract_inverted_index.without | 17 |
| abstract_inverted_index.However, | 28 |
| abstract_inverted_index.Learning | 59 |
| abstract_inverted_index.advanced | 137 |
| abstract_inverted_index.approach | 134 |
| abstract_inverted_index.editing, | 163 |
| abstract_inverted_index.editing. | 130 |
| abstract_inverted_index.existing | 29 |
| abstract_inverted_index.features | 86 |
| abstract_inverted_index.focusing | 64 |
| abstract_inverted_index.garnered | 25 |
| abstract_inverted_index.impeding | 38 |
| abstract_inverted_index.in-scope | 106 |
| abstract_inverted_index.inherent | 48 |
| abstract_inverted_index.inspired | 75 |
| abstract_inverted_index.language | 14 |
| abstract_inverted_index.parallel | 101 |
| abstract_inverted_index.process: | 89 |
| abstract_inverted_index.teaching | 66 |
| abstract_inverted_index.Alignment | 92 |
| abstract_inverted_index.Inference | 118 |
| abstract_inverted_index.Knowledge | 0 |
| abstract_inverted_index.answering | 51 |
| abstract_inverted_index.available | 178 |
| abstract_inverted_index.baselines | 138 |
| abstract_inverted_index.combining | 42 |
| abstract_inverted_index.comparing | 132 |
| abstract_inverted_index.impacting | 19 |
| abstract_inverted_index.knowledge | 11, 45, 49, 71, 129, 142, 154 |
| abstract_inverted_index.mechanism | 124 |
| abstract_inverted_index.real-time | 126 |
| abstract_inverted_index.reliable, | 105 |
| abstract_inverted_index.two-phase | 88 |
| abstract_inverted_index.attention. | 27 |
| abstract_inverted_index.benchmarks | 144 |
| abstract_inverted_index.fine-tunes | 95 |
| abstract_inverted_index.framework, | 63 |
| abstract_inverted_index.knowledge, | 37 |
| abstract_inverted_index.linguistic | 113 |
| abstract_inverted_index.memorizing | 34 |
| abstract_inverted_index.negatively | 18 |
| abstract_inverted_index.philosophy | 78 |
| abstract_inverted_index.preserving | 109 |
| abstract_inverted_index.proportion | 9 |
| abstract_inverted_index.questions, | 74 |
| abstract_inverted_index.questions. | 52 |
| abstract_inverted_index.robustness | 157 |
| abstract_inverted_index.sequential | 162 |
| abstract_inverted_index.widespread | 26 |
| abstract_inverted_index.demonstrate | 150 |
| abstract_inverted_index.effectively | 41 |
| abstract_inverted_index.efficiently | 5 |
| abstract_inverted_index.information | 111 |
| abstract_inverted_index.performance | 20 |
| abstract_inverted_index.superiority | 152 |
| abstract_inverted_index.techniques, | 2 |
| abstract_inverted_index.interference | 165 |
| abstract_inverted_index.meticulously | 99 |
| abstract_inverted_index.out-of-scope | 110 |
| abstract_inverted_index.performance, | 156 |
| abstract_inverted_index.proficiency; | 114 |
| abstract_inverted_index.predominantly | 31 |
| abstract_inverted_index.architectures, | 148 |
| abstract_inverted_index.retrieval-based | 123 |
| abstract_inverted_index.https://github.com/YJiangcm/LTE. | 180 |
| cited_by_percentile_year | |
| countries_distinct_count | 0 |
| institutions_distinct_count | 12 |
| citation_normalized_percentile |