{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:MULZLS7HDBBOPG427ZX6RDYG3O","short_pith_number":"pith:MULZLS7H","schema_version":"1.0","canonical_sha256":"651795cbe71842e79b9afe6fe88f06db8dddf7744da7622826199f11d317a364","source":{"kind":"arxiv","id":"2607.26873","version":1},"attestation_state":"computed","paper":{"title":"SERPO: Self-Evolving Rubric Policy Optimization for Open-Ended Test-Time Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Hua Yang, Jianze Wang, Jinlong Chen, Kunwang Zheng, Qianglong Chen, Qilong Zhang, Ying Liu, Yu Cao","submitted_at":"2026-07-29T13:03:07Z","abstract_excerpt":"Test-time reinforcement learning (TTRL) enables language models to self-evolve at inference time without labeled feedback. Existing methods rely on answer voting and therefore do not extend naturally to open-ended generation, where valid responses cannot be mapped to a shared canonical answer. Without external reward models or stronger judges, adaptation must instead construct reliable rewards from the model's own outputs. We introduce SERPO (Self-Evolving Rubric Policy Optimization), which replaces answer voting with a closed loop that co-evolves response evidence, query-specific rubrics, and"},"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":"2607.26873","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-07-29T13:03:07Z","cross_cats_sorted":[],"title_canon_sha256":"c04eab36a0bd3750bf7c81d4d163f8bf5f8cbbd6e9b68cda19d16a8801543202","abstract_canon_sha256":"3040e19887fabb366990990589c2ca8271a3a20c1764901eeb3fded84b189eb0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"651795cbe71842e79b9afe6fe88f06db8dddf7744da7622826199f11d317a364","last_reissued_at":"2026-07-30T01:22:50.354964Z","signature_status":"unsigned_v0","first_computed_at":"2026-07-30T01:22:50.354964Z"},"graph_snapshot":{"paper":{"title":"SERPO: Self-Evolving Rubric Policy Optimization for Open-Ended Test-Time Reinforcement Learning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Hua Yang, Jianze Wang, Jinlong Chen, Kunwang Zheng, Qianglong Chen, Qilong Zhang, Ying Liu, Yu Cao","submitted_at":"2026-07-29T13:03:07Z","abstract_excerpt":"Test-time reinforcement learning (TTRL) enables language models to self-evolve at inference time without labeled feedback. Existing methods rely on answer voting and therefore do not extend naturally to open-ended generation, where valid responses cannot be mapped to a shared canonical answer. Without external reward models or stronger judges, adaptation must instead construct reliable rewards from the model's own outputs. We introduce SERPO (Self-Evolving Rubric Policy Optimization), which replaces answer voting with a closed loop that co-evolves response evidence, query-specific rubrics, and"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.26873","kind":"arxiv","version":1},"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/2607.26873/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":"2607.26873","created_at":"2026-07-30T01:22:50.360029+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.26873v1","created_at":"2026-07-30T01:22:50.360029+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.26873","created_at":"2026-07-30T01:22:50.360029+00:00"},{"alias_kind":"pith_short_12","alias_value":"MULZLS7HDBBO","created_at":"2026-07-30T01:22:50.360029+00:00"},{"alias_kind":"pith_short_16","alias_value":"MULZLS7HDBBOPG42","created_at":"2026-07-30T01:22:50.360029+00:00"},{"alias_kind":"pith_short_8","alias_value":"MULZLS7H","created_at":"2026-07-30T01:22:50.360029+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MULZLS7HDBBOPG427ZX6RDYG3O","json":"https://pith.science/pith/MULZLS7HDBBOPG427ZX6RDYG3O.json","graph_json":"https://pith.science/api/pith-number/MULZLS7HDBBOPG427ZX6RDYG3O/graph.json","events_json":"https://pith.science/api/pith-number/MULZLS7HDBBOPG427ZX6RDYG3O/events.json","paper":"https://pith.science/paper/MULZLS7H"},"agent_actions":{"view_html":"https://pith.science/pith/MULZLS7HDBBOPG427ZX6RDYG3O","download_json":"https://pith.science/pith/MULZLS7HDBBOPG427ZX6RDYG3O.json","view_paper":"https://pith.science/paper/MULZLS7H","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.26873&json=true","fetch_graph":"https://pith.science/api/pith-number/MULZLS7HDBBOPG427ZX6RDYG3O/graph.json","fetch_events":"https://pith.science/api/pith-number/MULZLS7HDBBOPG427ZX6RDYG3O/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MULZLS7HDBBOPG427ZX6RDYG3O/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MULZLS7HDBBOPG427ZX6RDYG3O/action/storage_attestation","attest_author":"https://pith.science/pith/MULZLS7HDBBOPG427ZX6RDYG3O/action/author_attestation","sign_citation":"https://pith.science/pith/MULZLS7HDBBOPG427ZX6RDYG3O/action/citation_signature","submit_replication":"https://pith.science/pith/MULZLS7HDBBOPG427ZX6RDYG3O/action/replication_record"}},"created_at":"2026-07-30T01:22:50.360029+00:00","updated_at":"2026-07-30T01:22:50.360029+00:00"}