{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:257IWD2CAEQ3K7LNP6T3JITD2M","short_pith_number":"pith:257IWD2C","schema_version":"1.0","canonical_sha256":"d77e8b0f420121b57d6d7fa7b4a263d32fc074f13b390fc73e8b5b36820c57eb","source":{"kind":"arxiv","id":"2402.04624","version":2},"attestation_state":"computed","paper":{"title":"MEMORYLLM: Towards Self-Updatable Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Bing Yin, Haoming Jiang, Jingbo Shang, Jingfeng Yang, Julian McAuley, Qingyu Yin, Shiyang Li, Xian Li, Xiusi Chen, Yifan Gao, Yu Wang, Zheng Li","submitted_at":"2024-02-07T07:14:11Z","abstract_excerpt":"Existing Large Language Models (LLMs) usually remain static after deployment, which might make it hard to inject new knowledge into the model. We aim to build models containing a considerable portion of self-updatable parameters, enabling the model to integrate new knowledge effectively and efficiently. To this end, we introduce MEMORYLLM, a model that comprises a transformer and a fixed-size memory pool within the latent space of the transformer. MEMORYLLM can self-update with text knowledge and memorize the knowledge injected earlier. Our evaluations demonstrate the ability of MEMORYLLM to e"},"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":"2402.04624","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-02-07T07:14:11Z","cross_cats_sorted":[],"title_canon_sha256":"dc618bb424a8d33a089e9f078e7c87142c6845719a295ab143fc7558071ddbc7","abstract_canon_sha256":"6376243f87d00f912407c56da709fbc74171b2d4ebeb334cc60f93e5e7b82075"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:23:17.104486Z","signature_b64":"g/7B+bmGUuL8jDXoCxJWc+8v9ibU37VkTkwrapr50mz2zv/PrJKfYONVtYb4pJRxG6uyb+oLeYwAATlwzw6hCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d77e8b0f420121b57d6d7fa7b4a263d32fc074f13b390fc73e8b5b36820c57eb","last_reissued_at":"2026-07-05T08:23:17.103953Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:23:17.103953Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MEMORYLLM: Towards Self-Updatable Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Bing Yin, Haoming Jiang, Jingbo Shang, Jingfeng Yang, Julian McAuley, Qingyu Yin, Shiyang Li, Xian Li, Xiusi Chen, Yifan Gao, Yu Wang, Zheng Li","submitted_at":"2024-02-07T07:14:11Z","abstract_excerpt":"Existing Large Language Models (LLMs) usually remain static after deployment, which might make it hard to inject new knowledge into the model. We aim to build models containing a considerable portion of self-updatable parameters, enabling the model to integrate new knowledge effectively and efficiently. To this end, we introduce MEMORYLLM, a model that comprises a transformer and a fixed-size memory pool within the latent space of the transformer. MEMORYLLM can self-update with text knowledge and memorize the knowledge injected earlier. Our evaluations demonstrate the ability of MEMORYLLM to e"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.04624","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/2402.04624/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":"2402.04624","created_at":"2026-07-05T08:23:17.104030+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.04624v2","created_at":"2026-07-05T08:23:17.104030+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.04624","created_at":"2026-07-05T08:23:17.104030+00:00"},{"alias_kind":"pith_short_12","alias_value":"257IWD2CAEQ3","created_at":"2026-07-05T08:23:17.104030+00:00"},{"alias_kind":"pith_short_16","alias_value":"257IWD2CAEQ3K7LN","created_at":"2026-07-05T08:23:17.104030+00:00"},{"alias_kind":"pith_short_8","alias_value":"257IWD2C","created_at":"2026-07-05T08:23:17.104030+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":15,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.08032","citing_title":"What to Keep, What to Forget: A Rate--Distortion View of Memory Compaction in LLMs and Agents","ref_index":120,"is_internal_anchor":true},{"citing_arxiv_id":"2606.25161","citing_title":"TRUSTMEM: Learning Trustworthy Memory Consolidation for LLM Agents with Long-Term Memory","ref_index":30,"is_internal_anchor":false},{"citing_arxiv_id":"2606.20295","citing_title":"Token-Operations-Oriented Inference Optimization Techniques for Large Models","ref_index":111,"is_internal_anchor":false},{"citing_arxiv_id":"2606.17328","citing_title":"MemTrace: Probing What Final Accuracy Misses in Long-Term Memory","ref_index":50,"is_internal_anchor":false},{"citing_arxiv_id":"2607.01935","citing_title":"A-TMA: Decoupling State-Aware Memory Failures in Long-Term Agent Memory","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2606.05894","citing_title":"EMBER: Efficient Memory via Budgeted Evidence Retention for Long-Horizon Agents","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2606.30949","citing_title":"AgRefactor: Self-Evolving Agentic Workflow for HLS Compatibility and Performance","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2605.25680","citing_title":"Simulating Human Memory with Language Models","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2605.28773","citing_title":"Rethinking Memory as Continuously Evolving Connectivity","ref_index":51,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01223","citing_title":"Connecting the Dots: Benchmarking Reflective Memory in Long-Horizon Dialogue","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2511.02805","citing_title":"MemSearcher: Training LLMs to Reason, Search and Manage Memory via End-to-End Reinforcement Learning","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2504.15965","citing_title":"From Human Memory to AI Memory: A Survey on Memory Mechanisms in the Era of LLMs","ref_index":148,"is_internal_anchor":false},{"citing_arxiv_id":"2507.05257","citing_title":"Evaluating Memory in LLM Agents via Incremental Multi-Turn Interactions","ref_index":38,"is_internal_anchor":false},{"citing_arxiv_id":"2602.13933","citing_title":"HyMem: Hybrid Memory Architecture with Dynamic Retrieval Scheduling","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2604.06169","citing_title":"In-Place Test-Time Training","ref_index":56,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/257IWD2CAEQ3K7LNP6T3JITD2M","json":"https://pith.science/pith/257IWD2CAEQ3K7LNP6T3JITD2M.json","graph_json":"https://pith.science/api/pith-number/257IWD2CAEQ3K7LNP6T3JITD2M/graph.json","events_json":"https://pith.science/api/pith-number/257IWD2CAEQ3K7LNP6T3JITD2M/events.json","paper":"https://pith.science/paper/257IWD2C"},"agent_actions":{"view_html":"https://pith.science/pith/257IWD2CAEQ3K7LNP6T3JITD2M","download_json":"https://pith.science/pith/257IWD2CAEQ3K7LNP6T3JITD2M.json","view_paper":"https://pith.science/paper/257IWD2C","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.04624&json=true","fetch_graph":"https://pith.science/api/pith-number/257IWD2CAEQ3K7LNP6T3JITD2M/graph.json","fetch_events":"https://pith.science/api/pith-number/257IWD2CAEQ3K7LNP6T3JITD2M/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/257IWD2CAEQ3K7LNP6T3JITD2M/action/timestamp_anchor","attest_storage":"https://pith.science/pith/257IWD2CAEQ3K7LNP6T3JITD2M/action/storage_attestation","attest_author":"https://pith.science/pith/257IWD2CAEQ3K7LNP6T3JITD2M/action/author_attestation","sign_citation":"https://pith.science/pith/257IWD2CAEQ3K7LNP6T3JITD2M/action/citation_signature","submit_replication":"https://pith.science/pith/257IWD2CAEQ3K7LNP6T3JITD2M/action/replication_record"}},"created_at":"2026-07-05T08:23:17.104030+00:00","updated_at":"2026-07-05T08:23:17.104030+00:00"}