{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:Y6WZPXWMVRV6ZRNHD2KNGHT6FX","short_pith_number":"pith:Y6WZPXWM","schema_version":"1.0","canonical_sha256":"c7ad97deccac6becc5a71e94d31e7e2dde2bb8e9e1f0a44e443ecec85434b46e","source":{"kind":"arxiv","id":"2603.02218","version":2},"attestation_state":"computed","paper":{"title":"Self-Play Only Evolves When Self-Synthetic Pipeline Ensures Learnable Information Gain","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.IT","math.IT"],"primary_cat":"cs.LG","authors_text":"Siya Qi, Wei Liu, Yali Du, Yulan He","submitted_at":"2026-02-10T08:12:09Z","abstract_excerpt":"Large language models (LLMs) make it plausible to build systems that improve through self-evolving loops, but many existing proposals are better understood as self-play and often plateau quickly. A central failure mode is that the loop synthesises more data without increasing learnable information for the next iteration. Through experiments on a self-play coding task, we reveal that sustainable self-evolution requires a self-synthesised data pipeline with learnable information that increases across iterations. We identify triadic roles that self-evolving LLMs play: the Proposer, which generate"},"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":"2603.02218","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-02-10T08:12:09Z","cross_cats_sorted":["cs.AI","cs.CL","cs.IT","math.IT"],"title_canon_sha256":"55496a0b9781c4ebfd6631f51b8a50c976d679447d178bc90cf70b9f87f63f95","abstract_canon_sha256":"4bc666af11a3bb97cd65a65468a6313ee8651ff4eb7477fafafb127708f720ff"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:02:10.294115Z","signature_b64":"1NdDHoQhpOueQQz5jw7sMy7bGsqWpdrN9wwfWusCgMpTAQ4J+GFWrPBtm/cIWaAxJ1jRHysvcA7q0iOjByARAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c7ad97deccac6becc5a71e94d31e7e2dde2bb8e9e1f0a44e443ecec85434b46e","last_reissued_at":"2026-05-20T00:02:10.293380Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:02:10.293380Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Self-Play Only Evolves When Self-Synthetic Pipeline Ensures Learnable Information Gain","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.IT","math.IT"],"primary_cat":"cs.LG","authors_text":"Siya Qi, Wei Liu, Yali Du, Yulan He","submitted_at":"2026-02-10T08:12:09Z","abstract_excerpt":"Large language models (LLMs) make it plausible to build systems that improve through self-evolving loops, but many existing proposals are better understood as self-play and often plateau quickly. A central failure mode is that the loop synthesises more data without increasing learnable information for the next iteration. Through experiments on a self-play coding task, we reveal that sustainable self-evolution requires a self-synthesised data pipeline with learnable information that increases across iterations. We identify triadic roles that self-evolving LLMs play: the Proposer, which generate"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2603.02218","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/2603.02218/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":"2603.02218","created_at":"2026-05-20T00:02:10.293523+00:00"},{"alias_kind":"arxiv_version","alias_value":"2603.02218v2","created_at":"2026-05-20T00:02:10.293523+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2603.02218","created_at":"2026-05-20T00:02:10.293523+00:00"},{"alias_kind":"pith_short_12","alias_value":"Y6WZPXWMVRV6","created_at":"2026-05-20T00:02:10.293523+00:00"},{"alias_kind":"pith_short_16","alias_value":"Y6WZPXWMVRV6ZRNH","created_at":"2026-05-20T00:02:10.293523+00:00"},{"alias_kind":"pith_short_8","alias_value":"Y6WZPXWM","created_at":"2026-05-20T00:02:10.293523+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":4,"sample":[{"citing_arxiv_id":"2605.20086","citing_title":"What Do Evolutionary Coding Agents Evolve?","ref_index":31,"is_internal_anchor":true},{"citing_arxiv_id":"2604.03472","citing_title":"Vocabulary Dropout for Curriculum Diversity in LLM Co-Evolution","ref_index":22,"is_internal_anchor":true},{"citing_arxiv_id":"2604.17658","citing_title":"Towards Self-Improving Error Diagnosis in Multi-Agent Systems","ref_index":63,"is_internal_anchor":true},{"citing_arxiv_id":"2604.18320","citing_title":"EVE: Verifiable Self-Evolution of MLLMs via Executable Visual Transformations","ref_index":27,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/Y6WZPXWMVRV6ZRNHD2KNGHT6FX","json":"https://pith.science/pith/Y6WZPXWMVRV6ZRNHD2KNGHT6FX.json","graph_json":"https://pith.science/api/pith-number/Y6WZPXWMVRV6ZRNHD2KNGHT6FX/graph.json","events_json":"https://pith.science/api/pith-number/Y6WZPXWMVRV6ZRNHD2KNGHT6FX/events.json","paper":"https://pith.science/paper/Y6WZPXWM"},"agent_actions":{"view_html":"https://pith.science/pith/Y6WZPXWMVRV6ZRNHD2KNGHT6FX","download_json":"https://pith.science/pith/Y6WZPXWMVRV6ZRNHD2KNGHT6FX.json","view_paper":"https://pith.science/paper/Y6WZPXWM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2603.02218&json=true","fetch_graph":"https://pith.science/api/pith-number/Y6WZPXWMVRV6ZRNHD2KNGHT6FX/graph.json","fetch_events":"https://pith.science/api/pith-number/Y6WZPXWMVRV6ZRNHD2KNGHT6FX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/Y6WZPXWMVRV6ZRNHD2KNGHT6FX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/Y6WZPXWMVRV6ZRNHD2KNGHT6FX/action/storage_attestation","attest_author":"https://pith.science/pith/Y6WZPXWMVRV6ZRNHD2KNGHT6FX/action/author_attestation","sign_citation":"https://pith.science/pith/Y6WZPXWMVRV6ZRNHD2KNGHT6FX/action/citation_signature","submit_replication":"https://pith.science/pith/Y6WZPXWMVRV6ZRNHD2KNGHT6FX/action/replication_record"}},"created_at":"2026-05-20T00:02:10.293523+00:00","updated_at":"2026-05-20T00:02:10.293523+00:00"}