{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:24ZH7CZOHOCKQI5UBKEGORMGTF","short_pith_number":"pith:24ZH7CZO","schema_version":"1.0","canonical_sha256":"d7327f8b2e3b84a823b40a88674586996f0819a993e2e14038a9405da7ba7edf","source":{"kind":"arxiv","id":"2605.13641","version":1},"attestation_state":"computed","paper":{"title":"Multi-Objective and Mixed-Reward Reinforcement Learning via Reward-Decorrelated Policy Optimization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"RDPO stabilizes advantages in mixed-reward reinforcement learning by normalizing magnitudes and removing correlations before aggregation.","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Jiahong Zhou, Jingang Wang, Kaiyuan Liu, Rongxiang Weng, Xin Chen, Xunliang Cai, Yang Bai, Ziyuan Zhuang","submitted_at":"2026-05-13T15:05:18Z","abstract_excerpt":"Complex reinforcement learning environments frequently employ multi-task and mixed-reward formulations. In these settings, heterogeneous reward distributions and correlated reward dimensions often destabilize the construction of scalar advantages. To address these challenges, we propose Reward-Decorrelated Policy Optimization (RDPO), a reward-processing method designed to explicitly target both failure modes. RDPO first utilizes Magnitude-Aware Quantile normalization to stabilize prompt-level advantage allocation across binary, fractional, and continuous rewards. It then applies Mahalanobis wh"},"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":true,"formal_links_present":false},"canonical_record":{"source":{"id":"2605.13641","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-13T15:05:18Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"25c79d2d6d192456412abfcc88c8bcbce2ae575bf3285c8a57a47d72c4c76ca3","abstract_canon_sha256":"89ad6306ebae2e291fe2e899775f6b3612ffad186d089f93b57cdd1eaf0ce81c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T02:44:17.604038Z","signature_b64":"ezvoGmnOaWucUxs5maX02qkCX6LWkdPdWQm+/qtpyUafdnTVUzBDTtHZGdW+s0t9tgH7GzENJYcs85beLfDrBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d7327f8b2e3b84a823b40a88674586996f0819a993e2e14038a9405da7ba7edf","last_reissued_at":"2026-05-18T02:44:17.603486Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T02:44:17.603486Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Multi-Objective and Mixed-Reward Reinforcement Learning via Reward-Decorrelated Policy Optimization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"RDPO stabilizes advantages in mixed-reward reinforcement learning by normalizing magnitudes and removing correlations before aggregation.","cross_cats":["cs.CL"],"primary_cat":"cs.LG","authors_text":"Jiahong Zhou, Jingang Wang, Kaiyuan Liu, Rongxiang Weng, Xin Chen, Xunliang Cai, Yang Bai, Ziyuan Zhuang","submitted_at":"2026-05-13T15:05:18Z","abstract_excerpt":"Complex reinforcement learning environments frequently employ multi-task and mixed-reward formulations. In these settings, heterogeneous reward distributions and correlated reward dimensions often destabilize the construction of scalar advantages. To address these challenges, we propose Reward-Decorrelated Policy Optimization (RDPO), a reward-processing method designed to explicitly target both failure modes. RDPO first utilizes Magnitude-Aware Quantile normalization to stabilize prompt-level advantage allocation across binary, fractional, and continuous rewards. It then applies Mahalanobis wh"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"When applied during the post-training of LongCat-Flash, RDPO enhances instruction following, writing quality, and robustness to hard prompts while remaining broadly competitive on reasoning and coding evaluations.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That magnitude-aware quantile normalization and Mahalanobis whitening will stabilize advantages across heterogeneous rewards without discarding critical signal or introducing new biases in the specific reward distributions of the LongCat-Flash training setup.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"RDPO applies magnitude-aware quantile normalization and Mahalanobis whitening to decorrelate heterogeneous rewards in multi-objective RL, improving instruction following and writing quality on LongCat-Flash post-training while staying competitive on reasoning and coding.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"RDPO stabilizes advantages in mixed-reward reinforcement learning by normalizing magnitudes and removing correlations before aggregation.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"972ae1a76c20a2b7956de84521d7621a62b7068bf1f85c8e45fa9f265b7730a4"},"source":{"id":"2605.13641","kind":"arxiv","version":1},"verdict":{"id":"31a9043e-c6f2-4445-b53d-1950b87a7d90","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T19:20:34.498141Z","strongest_claim":"When applied during the post-training of LongCat-Flash, RDPO enhances instruction following, writing quality, and robustness to hard prompts while remaining broadly competitive on reasoning and coding evaluations.","one_line_summary":"RDPO applies magnitude-aware quantile normalization and Mahalanobis whitening to decorrelate heterogeneous rewards in multi-objective RL, improving instruction following and writing quality on LongCat-Flash post-training while staying competitive on reasoning and coding.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That magnitude-aware quantile normalization and Mahalanobis whitening will stabilize advantages across heterogeneous rewards without discarding critical signal or introducing new biases in the specific reward distributions of the LongCat-Flash training setup.","pith_extraction_headline":"RDPO stabilizes advantages in mixed-reward reinforcement learning by normalizing magnitudes and removing correlations before aggregation."},"references":{"count":17,"sample":[{"doi":"","year":null,"title":"DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models","work_id":"c5006563-f3ec-438a-9e35-b7b484f34828","ref_index":1,"cited_arxiv_id":"2402.03300","is_internal_anchor":true},{"doi":"10.48550/arxiv.2402.03300","year":null,"title":"DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models","work_id":"c5006563-f3ec-438a-9e35-b7b484f34828","ref_index":2,"cited_arxiv_id":"2402.03300","is_internal_anchor":true},{"doi":"10.48550/arxiv.2601.05242","year":null,"title":"GDPO: Group reward-Decoupled Normalization Policy Optimization for Multi-reward RL Optimization","work_id":"e60e0bf8-33b8-47c6-8d66-7975f5c121d4","ref_index":3,"cited_arxiv_id":"2601.05242","is_internal_anchor":true},{"doi":"10.48550/arxiv.2501.12599","year":null,"title":"Kimi k1.5: Scaling Reinforcement Learning with LLMs","work_id":"bff96ab1-bd6a-4585-be23-74fdb51969c7","ref_index":4,"cited_arxiv_id":"2501.12599","is_internal_anchor":true},{"doi":"10.48550/arxiv.2503.04697","year":null,"title":"L1: Controlling How Long A Reasoning Model Thinks With Reinforcement Learning","work_id":"ad7236fb-3752-48a9-b782-a384899c45a0","ref_index":5,"cited_arxiv_id":"2503.04697","is_internal_anchor":true}],"resolved_work":17,"snapshot_sha256":"1e5b013d5cda89c960531d62204d94226a0666c2a6d5024d24bb00b4b409fcbf","internal_anchors":9},"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":"2605.13641","created_at":"2026-05-18T02:44:17.603588+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.13641v1","created_at":"2026-05-18T02:44:17.603588+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.13641","created_at":"2026-05-18T02:44:17.603588+00:00"},{"alias_kind":"pith_short_12","alias_value":"24ZH7CZOHOCK","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_16","alias_value":"24ZH7CZOHOCKQI5U","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_8","alias_value":"24ZH7CZO","created_at":"2026-05-18T12:33:37.589309+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/24ZH7CZOHOCKQI5UBKEGORMGTF","json":"https://pith.science/pith/24ZH7CZOHOCKQI5UBKEGORMGTF.json","graph_json":"https://pith.science/api/pith-number/24ZH7CZOHOCKQI5UBKEGORMGTF/graph.json","events_json":"https://pith.science/api/pith-number/24ZH7CZOHOCKQI5UBKEGORMGTF/events.json","paper":"https://pith.science/paper/24ZH7CZO"},"agent_actions":{"view_html":"https://pith.science/pith/24ZH7CZOHOCKQI5UBKEGORMGTF","download_json":"https://pith.science/pith/24ZH7CZOHOCKQI5UBKEGORMGTF.json","view_paper":"https://pith.science/paper/24ZH7CZO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.13641&json=true","fetch_graph":"https://pith.science/api/pith-number/24ZH7CZOHOCKQI5UBKEGORMGTF/graph.json","fetch_events":"https://pith.science/api/pith-number/24ZH7CZOHOCKQI5UBKEGORMGTF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/24ZH7CZOHOCKQI5UBKEGORMGTF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/24ZH7CZOHOCKQI5UBKEGORMGTF/action/storage_attestation","attest_author":"https://pith.science/pith/24ZH7CZOHOCKQI5UBKEGORMGTF/action/author_attestation","sign_citation":"https://pith.science/pith/24ZH7CZOHOCKQI5UBKEGORMGTF/action/citation_signature","submit_replication":"https://pith.science/pith/24ZH7CZOHOCKQI5UBKEGORMGTF/action/replication_record"}},"created_at":"2026-05-18T02:44:17.603588+00:00","updated_at":"2026-05-18T02:44:17.603588+00:00"}