{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:YDJ2XWWTDX7XU5NG43HBEL6QIF","short_pith_number":"pith:YDJ2XWWT","schema_version":"1.0","canonical_sha256":"c0d3abdad31dff7a75a6e6ce122fd041555ab5da60aed16fc032174c4492339d","source":{"kind":"arxiv","id":"2603.11784","version":2},"attestation_state":"computed","paper":{"title":"Language Generation with Replay: A Learning-Theoretic View of Model Collapse","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Amartya Sanyal, Giorgio Racca, Michal Valko","submitted_at":"2026-03-12T10:44:17Z","abstract_excerpt":"As scaling laws push the training of frontier large language models (LLMs) toward ever-growing data requirements, training pipelines are approaching a regime where much of the publicly available online text may be consumed. At the same time, widespread LLM usage increases the volume of machine-generated content on the web; together, these trends raise the likelihood of generated text re-entering future training corpora, increasing the associated risk of performance degradation often called model collapse. In practice, model developers address this concern through data cleaning, watermarking, s"},"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":"2603.11784","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-03-12T10:44:17Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"0a82030e20bc1168f15106815a8f4b065cfb3338deb1f92be9f28c6beb8f45a5","abstract_canon_sha256":"c1a9711d8f7151b6a9bf6a7fda800ea4b9bcb8e5476a567c28568fbd6d4d96a6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-07T02:18:37.154936Z","signature_b64":"OXcAIKx1por9T2T64hA0ZOR1D9iGjtr9oi/c5Ndt2nK6RnR9SdtEtF1esAYxu2FW/ZWHypeHW7NjJ5S6RhzvBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c0d3abdad31dff7a75a6e6ce122fd041555ab5da60aed16fc032174c4492339d","last_reissued_at":"2026-07-07T02:18:37.153930Z","signature_status":"signed_v1","first_computed_at":"2026-07-07T02:18:37.153930Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Language Generation with Replay: A Learning-Theoretic View of Model Collapse","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Amartya Sanyal, Giorgio Racca, Michal Valko","submitted_at":"2026-03-12T10:44:17Z","abstract_excerpt":"As scaling laws push the training of frontier large language models (LLMs) toward ever-growing data requirements, training pipelines are approaching a regime where much of the publicly available online text may be consumed. At the same time, widespread LLM usage increases the volume of machine-generated content on the web; together, these trends raise the likelihood of generated text re-entering future training corpora, increasing the associated risk of performance degradation often called model collapse. In practice, model developers address this concern through data cleaning, watermarking, s"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2603.11784","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/2603.11784/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":"2603.11784","created_at":"2026-07-07T02:18:37.154108+00:00"},{"alias_kind":"arxiv_version","alias_value":"2603.11784v2","created_at":"2026-07-07T02:18:37.154108+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2603.11784","created_at":"2026-07-07T02:18:37.154108+00:00"},{"alias_kind":"pith_short_12","alias_value":"YDJ2XWWTDX7X","created_at":"2026-07-07T02:18:37.154108+00:00"},{"alias_kind":"pith_short_16","alias_value":"YDJ2XWWTDX7XU5NG","created_at":"2026-07-07T02:18:37.154108+00:00"},{"alias_kind":"pith_short_8","alias_value":"YDJ2XWWT","created_at":"2026-07-07T02:18:37.154108+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2608.05029","citing_title":"On Computational Hardness of Mistake-Bounded Language Generation: A Random-Oracle Query Separation","ref_index":26,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YDJ2XWWTDX7XU5NG43HBEL6QIF","json":"https://pith.science/pith/YDJ2XWWTDX7XU5NG43HBEL6QIF.json","graph_json":"https://pith.science/api/pith-number/YDJ2XWWTDX7XU5NG43HBEL6QIF/graph.json","events_json":"https://pith.science/api/pith-number/YDJ2XWWTDX7XU5NG43HBEL6QIF/events.json","paper":"https://pith.science/paper/YDJ2XWWT"},"agent_actions":{"view_html":"https://pith.science/pith/YDJ2XWWTDX7XU5NG43HBEL6QIF","download_json":"https://pith.science/pith/YDJ2XWWTDX7XU5NG43HBEL6QIF.json","view_paper":"https://pith.science/paper/YDJ2XWWT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2603.11784&json=true","fetch_graph":"https://pith.science/api/pith-number/YDJ2XWWTDX7XU5NG43HBEL6QIF/graph.json","fetch_events":"https://pith.science/api/pith-number/YDJ2XWWTDX7XU5NG43HBEL6QIF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YDJ2XWWTDX7XU5NG43HBEL6QIF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YDJ2XWWTDX7XU5NG43HBEL6QIF/action/storage_attestation","attest_author":"https://pith.science/pith/YDJ2XWWTDX7XU5NG43HBEL6QIF/action/author_attestation","sign_citation":"https://pith.science/pith/YDJ2XWWTDX7XU5NG43HBEL6QIF/action/citation_signature","submit_replication":"https://pith.science/pith/YDJ2XWWTDX7XU5NG43HBEL6QIF/action/replication_record"}},"created_at":"2026-07-07T02:18:37.154108+00:00","updated_at":"2026-07-07T02:18:37.154108+00:00"}