{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:LFRMKKPQBRP6IPUALV32KS4O5S","short_pith_number":"pith:LFRMKKPQ","schema_version":"1.0","canonical_sha256":"5962c529f00c5fe43e805d77a54b8eecbf0964220eb1111d96dc316d22da5467","source":{"kind":"arxiv","id":"2605.26958","version":1},"attestation_state":"computed","paper":{"title":"Tournament-GRPO: Group-Wise Tournament Rewards for Reinforcement Learning in Open-Ended Long-Form Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Erhan Zhang, Jiaxin Mao, Wei Yang, Xiaochi Wei, Yan Gao, Yao Hu, Yiqun Chen, Yi Wu, Zihan Shen, Zixuan Yang","submitted_at":"2026-05-26T12:49:19Z","abstract_excerpt":"Reinforcement learning in open-ended long-form generation is challenging because reliable reference answers and automatic metrics are often unavailable. Existing rubric-based methods typically rely on pointwise LLM-as-a-judge scoring, but absolute scores are difficult to calibrate across complex responses, may provide weak discrimination among same-query rollouts, and can become saturated during optimization. We propose Tournament-GRPO, a group-wise reward framework that converts rubric-guided LLM judgments into relative rewards through repeated multi-round tournaments among same-query rollout"},"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":"2605.26958","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-05-26T12:49:19Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"33c2bb95ba8845494b877346b1c04f33f00efd8d27c864a0b0a3636530afe2e5","abstract_canon_sha256":"37f11a86c8ca8a16cbf827cb99b982a7a98f6ec05b363f7dd44438b676a9adc1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-27T01:06:21.637866Z","signature_b64":"H0DCHLdT8lP3inw42HY3nfZePr3o+/cjaDBG5QQQFhISsvqxyQJzRsvsEDRX7pQDg1FBPBwGPR+XzX8X3FF0CQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5962c529f00c5fe43e805d77a54b8eecbf0964220eb1111d96dc316d22da5467","last_reissued_at":"2026-05-27T01:06:21.637077Z","signature_status":"signed_v1","first_computed_at":"2026-05-27T01:06:21.637077Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Tournament-GRPO: Group-Wise Tournament Rewards for Reinforcement Learning in Open-Ended Long-Form Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Erhan Zhang, Jiaxin Mao, Wei Yang, Xiaochi Wei, Yan Gao, Yao Hu, Yiqun Chen, Yi Wu, Zihan Shen, Zixuan Yang","submitted_at":"2026-05-26T12:49:19Z","abstract_excerpt":"Reinforcement learning in open-ended long-form generation is challenging because reliable reference answers and automatic metrics are often unavailable. Existing rubric-based methods typically rely on pointwise LLM-as-a-judge scoring, but absolute scores are difficult to calibrate across complex responses, may provide weak discrimination among same-query rollouts, and can become saturated during optimization. We propose Tournament-GRPO, a group-wise reward framework that converts rubric-guided LLM judgments into relative rewards through repeated multi-round tournaments among same-query rollout"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.26958","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/2605.26958/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":"2605.26958","created_at":"2026-05-27T01:06:21.637205+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.26958v1","created_at":"2026-05-27T01:06:21.637205+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.26958","created_at":"2026-05-27T01:06:21.637205+00:00"},{"alias_kind":"pith_short_12","alias_value":"LFRMKKPQBRP6","created_at":"2026-05-27T01:06:21.637205+00:00"},{"alias_kind":"pith_short_16","alias_value":"LFRMKKPQBRP6IPUA","created_at":"2026-05-27T01:06:21.637205+00:00"},{"alias_kind":"pith_short_8","alias_value":"LFRMKKPQ","created_at":"2026-05-27T01:06:21.637205+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/LFRMKKPQBRP6IPUALV32KS4O5S","json":"https://pith.science/pith/LFRMKKPQBRP6IPUALV32KS4O5S.json","graph_json":"https://pith.science/api/pith-number/LFRMKKPQBRP6IPUALV32KS4O5S/graph.json","events_json":"https://pith.science/api/pith-number/LFRMKKPQBRP6IPUALV32KS4O5S/events.json","paper":"https://pith.science/paper/LFRMKKPQ"},"agent_actions":{"view_html":"https://pith.science/pith/LFRMKKPQBRP6IPUALV32KS4O5S","download_json":"https://pith.science/pith/LFRMKKPQBRP6IPUALV32KS4O5S.json","view_paper":"https://pith.science/paper/LFRMKKPQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.26958&json=true","fetch_graph":"https://pith.science/api/pith-number/LFRMKKPQBRP6IPUALV32KS4O5S/graph.json","fetch_events":"https://pith.science/api/pith-number/LFRMKKPQBRP6IPUALV32KS4O5S/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LFRMKKPQBRP6IPUALV32KS4O5S/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LFRMKKPQBRP6IPUALV32KS4O5S/action/storage_attestation","attest_author":"https://pith.science/pith/LFRMKKPQBRP6IPUALV32KS4O5S/action/author_attestation","sign_citation":"https://pith.science/pith/LFRMKKPQBRP6IPUALV32KS4O5S/action/citation_signature","submit_replication":"https://pith.science/pith/LFRMKKPQBRP6IPUALV32KS4O5S/action/replication_record"}},"created_at":"2026-05-27T01:06:21.637205+00:00","updated_at":"2026-05-27T01:06:21.637205+00:00"}