{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:BZY4HYLFNUCUDOJER5JCL55Q5J","short_pith_number":"pith:BZY4HYLF","schema_version":"1.0","canonical_sha256":"0e71c3e1656d0541b9248f5225f7b0ea6570253f4c314e3c9323edf91232aec4","source":{"kind":"arxiv","id":"2410.12341","version":4},"attestation_state":"computed","paper":{"title":"Learning by Surprise: Adaptive Mitigation of Model Collapse in Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Alistair Knott, Daniele Gambetta, Dino Pedreschi, Fosca Giannotti, Gizem Gezici, Luca Pappalardo","submitted_at":"2024-10-16T08:02:48Z","abstract_excerpt":"As AI-generated content increasingly populates the web, generative AI models are at growing risk of being trained on their own outputs, a process known as AI autophagy. This feedback loop has been shown to induce model collapse, typically characterized by a loss of diversity in generated content. However, existing work offers a limited understanding of this phenomenon and relies on mitigation strategies that assume access to human-authored data. In this paper, we conduct extensive simulations across multiple datasets and LLMs to address key gaps in the study of model collapse. First, we introd"},"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":"2410.12341","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-10-16T08:02:48Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"ed80e96fd6911b031e337dfc67488b4a7a59bb7a7fbe870ca4f27711a380b619","abstract_canon_sha256":"10df019373c94f9141047c49d7c6782f339ad97bdd3b0030db271d2778d77f91"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-01T01:17:38.282148Z","signature_b64":"xROH4OvEQVmJwCQswbbFju3TXBsbqDcKhMqTZpnUD4B5dAykTdidJGlpOmaCdf4Y29bS34rYH3EX9a3veMZaAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0e71c3e1656d0541b9248f5225f7b0ea6570253f4c314e3c9323edf91232aec4","last_reissued_at":"2026-07-01T01:17:38.281468Z","signature_status":"signed_v1","first_computed_at":"2026-07-01T01:17:38.281468Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning by Surprise: Adaptive Mitigation of Model Collapse in Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Alistair Knott, Daniele Gambetta, Dino Pedreschi, Fosca Giannotti, Gizem Gezici, Luca Pappalardo","submitted_at":"2024-10-16T08:02:48Z","abstract_excerpt":"As AI-generated content increasingly populates the web, generative AI models are at growing risk of being trained on their own outputs, a process known as AI autophagy. This feedback loop has been shown to induce model collapse, typically characterized by a loss of diversity in generated content. However, existing work offers a limited understanding of this phenomenon and relies on mitigation strategies that assume access to human-authored data. In this paper, we conduct extensive simulations across multiple datasets and LLMs to address key gaps in the study of model collapse. First, we introd"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.12341","kind":"arxiv","version":4},"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/2410.12341/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":"2410.12341","created_at":"2026-07-01T01:17:38.281554+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.12341v4","created_at":"2026-07-01T01:17:38.281554+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.12341","created_at":"2026-07-01T01:17:38.281554+00:00"},{"alias_kind":"pith_short_12","alias_value":"BZY4HYLFNUCU","created_at":"2026-07-01T01:17:38.281554+00:00"},{"alias_kind":"pith_short_16","alias_value":"BZY4HYLFNUCUDOJE","created_at":"2026-07-01T01:17:38.281554+00:00"},{"alias_kind":"pith_short_8","alias_value":"BZY4HYLF","created_at":"2026-07-01T01:17:38.281554+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":3,"sample":[{"citing_arxiv_id":"2607.00032","citing_title":"The MMM Data Model -- A Normative Specification for Knowledge Interoperability in a Decentralisable Knowledge Commons","ref_index":24,"is_internal_anchor":true},{"citing_arxiv_id":"2606.02339","citing_title":"Entropy Minimization without Model Collapse: Mitigating Prediction Bias in Medical Imaging","ref_index":21,"is_internal_anchor":true},{"citing_arxiv_id":"2605.07724","citing_title":"Curated Synthetic Data Doesn't Have to Collapse: A Theoretical Study of Generative Retraining with Pluralistic Preferences","ref_index":92,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/BZY4HYLFNUCUDOJER5JCL55Q5J","json":"https://pith.science/pith/BZY4HYLFNUCUDOJER5JCL55Q5J.json","graph_json":"https://pith.science/api/pith-number/BZY4HYLFNUCUDOJER5JCL55Q5J/graph.json","events_json":"https://pith.science/api/pith-number/BZY4HYLFNUCUDOJER5JCL55Q5J/events.json","paper":"https://pith.science/paper/BZY4HYLF"},"agent_actions":{"view_html":"https://pith.science/pith/BZY4HYLFNUCUDOJER5JCL55Q5J","download_json":"https://pith.science/pith/BZY4HYLFNUCUDOJER5JCL55Q5J.json","view_paper":"https://pith.science/paper/BZY4HYLF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.12341&json=true","fetch_graph":"https://pith.science/api/pith-number/BZY4HYLFNUCUDOJER5JCL55Q5J/graph.json","fetch_events":"https://pith.science/api/pith-number/BZY4HYLFNUCUDOJER5JCL55Q5J/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BZY4HYLFNUCUDOJER5JCL55Q5J/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BZY4HYLFNUCUDOJER5JCL55Q5J/action/storage_attestation","attest_author":"https://pith.science/pith/BZY4HYLFNUCUDOJER5JCL55Q5J/action/author_attestation","sign_citation":"https://pith.science/pith/BZY4HYLFNUCUDOJER5JCL55Q5J/action/citation_signature","submit_replication":"https://pith.science/pith/BZY4HYLFNUCUDOJER5JCL55Q5J/action/replication_record"}},"created_at":"2026-07-01T01:17:38.281554+00:00","updated_at":"2026-07-01T01:17:38.281554+00:00"}