{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:YVC6JH55MDQELF7DJJP3ISGZMH","short_pith_number":"pith:YVC6JH55","schema_version":"1.0","canonical_sha256":"c545e49fbd60e04597e34a5fb448d961f69367bf0766be8083fe6d701625c560","source":{"kind":"arxiv","id":"2402.01364","version":2},"attestation_state":"computed","paper":{"title":"Continual Learning for Large Language Models: A Survey","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Gholamreza Haffari, Linhao Luo, Shirui Pan, Thuy-Trang Vu, Tongtong Wu, Yuan-Fang Li","submitted_at":"2024-02-02T12:34:09Z","abstract_excerpt":"Large language models (LLMs) are not amenable to frequent re-training, due to high training costs arising from their massive scale. However, updates are necessary to endow LLMs with new skills and keep them up-to-date with rapidly evolving human knowledge. This paper surveys recent works on continual learning for LLMs. Due to the unique nature of LLMs, we catalog continue learning techniques in a novel multi-staged categorization scheme, involving continual pretraining, instruction tuning, and alignment. We contrast continual learning for LLMs with simpler adaptation methods used in smaller mo"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2402.01364","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-02-02T12:34:09Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"2e3a2dedf65a4a13a44826a6a23c42fa6bbbd7c4312de4ff0d8fe434fb73d42d","abstract_canon_sha256":"0b702538048e3a139e11adb62cc6654cde10e92064a0a98330f1b9d0480c07f9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:42:23.731378Z","signature_b64":"xLDdDwk+Ml2hyTdC0+h2jyLMi+zF4QMyW9f4qcpMqRBgHyKPgjUxHwhuMOC+C5Ptsuxawo/Aeo2X8y1uQprsAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c545e49fbd60e04597e34a5fb448d961f69367bf0766be8083fe6d701625c560","last_reissued_at":"2026-07-05T07:42:23.730833Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:42:23.730833Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Continual Learning for Large Language Models: A Survey","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Gholamreza Haffari, Linhao Luo, Shirui Pan, Thuy-Trang Vu, Tongtong Wu, Yuan-Fang Li","submitted_at":"2024-02-02T12:34:09Z","abstract_excerpt":"Large language models (LLMs) are not amenable to frequent re-training, due to high training costs arising from their massive scale. However, updates are necessary to endow LLMs with new skills and keep them up-to-date with rapidly evolving human knowledge. This paper surveys recent works on continual learning for LLMs. Due to the unique nature of LLMs, we catalog continue learning techniques in a novel multi-staged categorization scheme, involving continual pretraining, instruction tuning, and alignment. We contrast continual learning for LLMs with simpler adaptation methods used in smaller mo"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.01364","kind":"arxiv","version":2},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2402.01364/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2402.01364","created_at":"2026-07-05T07:42:23.730918+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.01364v2","created_at":"2026-07-05T07:42:23.730918+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.01364","created_at":"2026-07-05T07:42:23.730918+00:00"},{"alias_kind":"pith_short_12","alias_value":"YVC6JH55MDQE","created_at":"2026-07-05T07:42:23.730918+00:00"},{"alias_kind":"pith_short_16","alias_value":"YVC6JH55MDQELF7D","created_at":"2026-07-05T07:42:23.730918+00:00"},{"alias_kind":"pith_short_8","alias_value":"YVC6JH55","created_at":"2026-07-05T07:42:23.730918+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":22,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.21089","citing_title":"Repeated post-training is not Self-improving: Diagnosing Scientific Amnesia in Continual DPO Pipelines","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2606.24901","citing_title":"LLM Evolution as an Industry-Scale Ecosystem: A Lifecycle Perspective on Continual Learning","ref_index":116,"is_internal_anchor":false},{"citing_arxiv_id":"2606.07500","citing_title":"Sparse Subspace-to-Expert Sharing for Task-Agnostic Continual Learning","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2606.06698","citing_title":"RECAP: Regression Evaluation for Continual Adaptation of Prompts","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2606.04703","citing_title":"Rethinking Continual Experience Internalization for Self-Evolving LLM Agents","ref_index":74,"is_internal_anchor":false},{"citing_arxiv_id":"2605.30857","citing_title":"MADS: Model-Aware Diverse Core Set Selection for Instruction Tuning","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2606.28900","citing_title":"MedEvoEval: Evaluating Continual Evolution of Doctor Agents through Simulated Clinical Episodes","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2605.26097","citing_title":"Forgetting in Language Models: Capacity, Optimization, and Self-Generated Replay","ref_index":48,"is_internal_anchor":false},{"citing_arxiv_id":"2411.04832","citing_title":"Plasticity Loss in Deep Reinforcement Learning: A Survey","ref_index":108,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15104","citing_title":"From Text to Voice: A Reproducible and Verifiable Framework for Evaluating Tool Calling LLM Agents","ref_index":43,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15156","citing_title":"MeMo: Memory as a Model","ref_index":82,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17361","citing_title":"MasFACT: Continual Multi-Agent Topology Learning via Geometry-Aware Posterior Transfer","ref_index":42,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15384","citing_title":"Is One Score Enough? Rethinking the Evaluation of Sequentially Evolving LLM Memory","ref_index":38,"is_internal_anchor":false},{"citing_arxiv_id":"2507.03617","citing_title":"EMERGE: A Benchmark for Updating Knowledge Graphs with Emerging Textual Knowledge","ref_index":70,"is_internal_anchor":false},{"citing_arxiv_id":"2509.17183","citing_title":"LifeAlign: Lifelong Alignment for Large Language Models with Memory-Augmented Focalized Preference Optimization","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2603.13683","citing_title":"Preconditioned Test-Time Adaptation for Out-of-Distribution Debiasing in Narrative Generation","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15156","citing_title":"MeMo: Memory as a Model","ref_index":81,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10765","citing_title":"Dynamic Cross-Modal Prompt Generation for Multimodal Continual Instruction Tuning","ref_index":43,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10391","citing_title":"Phoenix-VL 1.5 Medium Technical Report","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2412.05579","citing_title":"LLMs-as-Judges: A Comprehensive Survey on LLM-based Evaluation Methods","ref_index":253,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05495","citing_title":"Shortcut Solutions Learned by Transformers Impair Continual Compositional Reasoning","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2604.14718","citing_title":"The Agentification of Scientific Research: A Physicist's Perspective","ref_index":31,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YVC6JH55MDQELF7DJJP3ISGZMH","json":"https://pith.science/pith/YVC6JH55MDQELF7DJJP3ISGZMH.json","graph_json":"https://pith.science/api/pith-number/YVC6JH55MDQELF7DJJP3ISGZMH/graph.json","events_json":"https://pith.science/api/pith-number/YVC6JH55MDQELF7DJJP3ISGZMH/events.json","paper":"https://pith.science/paper/YVC6JH55"},"agent_actions":{"view_html":"https://pith.science/pith/YVC6JH55MDQELF7DJJP3ISGZMH","download_json":"https://pith.science/pith/YVC6JH55MDQELF7DJJP3ISGZMH.json","view_paper":"https://pith.science/paper/YVC6JH55","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.01364&json=true","fetch_graph":"https://pith.science/api/pith-number/YVC6JH55MDQELF7DJJP3ISGZMH/graph.json","fetch_events":"https://pith.science/api/pith-number/YVC6JH55MDQELF7DJJP3ISGZMH/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YVC6JH55MDQELF7DJJP3ISGZMH/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YVC6JH55MDQELF7DJJP3ISGZMH/action/storage_attestation","attest_author":"https://pith.science/pith/YVC6JH55MDQELF7DJJP3ISGZMH/action/author_attestation","sign_citation":"https://pith.science/pith/YVC6JH55MDQELF7DJJP3ISGZMH/action/citation_signature","submit_replication":"https://pith.science/pith/YVC6JH55MDQELF7DJJP3ISGZMH/action/replication_record"}},"created_at":"2026-07-05T07:42:23.730918+00:00","updated_at":"2026-07-05T07:42:23.730918+00:00"}