{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:Q2IDQ7FXI4Y5WGXEOBISVZDX52","short_pith_number":"pith:Q2IDQ7FX","schema_version":"1.0","canonical_sha256":"8690387cb74731db1ae470512ae477ee8b377c5306747ac7a3a61769b965797b","source":{"kind":"arxiv","id":"2607.16169","version":1},"attestation_state":"computed","paper":{"title":"When Does Muon Help Agentic Reinforcement Learning?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Hao Sun, Jinghao Lin, Kai Ruan, Qianshan Wei, Xuan Wang, Zihe Huang, Ziqi Zhou","submitted_at":"2026-07-17T17:49:05Z","abstract_excerpt":"Muon is competitive with AdamW in large-scale pre-training, but its value for reinforcement-learning (RL) post-training remains unclear. We study vanilla Muon in sparse-reward agentic RL through matched single-seed comparisons with AdamW on ALFWorld using Qwen2.5-0.5B-Instruct. Under Group-in-Group Policy Optimization (GiGPO), applying Muon only to hidden weight matrices raises final-window validation success from 0.290 to 0.546 (+88%); high-rate AdamW controls retain no post-update success. The effect depends on the advantage estimator and learning rate. At 3e-5, Muon improves GRPO from 0.161"},"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":"2607.16169","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-07-17T17:49:05Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"43722afa4d026a0adf18653612cc4360bb8a57acd5b63210c00e0e4b8d39d167","abstract_canon_sha256":"1531bdfee6a0c0077b5d134abfee066db4a470d2a6d307712f1a1125a374835c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-20T02:19:37.361715Z","signature_b64":"eM4cEt1vE2v9BiJa5ugNayV0fvaWo4Uw9ZGn9LJ+zisiNAtyrDp6h4t5dEWcx2OSSW5c+LSqBJivqRVXsEpIDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8690387cb74731db1ae470512ae477ee8b377c5306747ac7a3a61769b965797b","last_reissued_at":"2026-07-20T02:19:37.360025Z","signature_status":"signed_v1","first_computed_at":"2026-07-20T02:19:37.360025Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"When Does Muon Help Agentic Reinforcement Learning?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Hao Sun, Jinghao Lin, Kai Ruan, Qianshan Wei, Xuan Wang, Zihe Huang, Ziqi Zhou","submitted_at":"2026-07-17T17:49:05Z","abstract_excerpt":"Muon is competitive with AdamW in large-scale pre-training, but its value for reinforcement-learning (RL) post-training remains unclear. We study vanilla Muon in sparse-reward agentic RL through matched single-seed comparisons with AdamW on ALFWorld using Qwen2.5-0.5B-Instruct. Under Group-in-Group Policy Optimization (GiGPO), applying Muon only to hidden weight matrices raises final-window validation success from 0.290 to 0.546 (+88%); high-rate AdamW controls retain no post-update success. The effect depends on the advantage estimator and learning rate. At 3e-5, Muon improves GRPO from 0.161"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.16169","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/2607.16169/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":"2607.16169","created_at":"2026-07-20T02:19:37.361257+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.16169v1","created_at":"2026-07-20T02:19:37.361257+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.16169","created_at":"2026-07-20T02:19:37.361257+00:00"},{"alias_kind":"pith_short_12","alias_value":"Q2IDQ7FXI4Y5","created_at":"2026-07-20T02:19:37.361257+00:00"},{"alias_kind":"pith_short_16","alias_value":"Q2IDQ7FXI4Y5WGXE","created_at":"2026-07-20T02:19:37.361257+00:00"},{"alias_kind":"pith_short_8","alias_value":"Q2IDQ7FX","created_at":"2026-07-20T02:19:37.361257+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/Q2IDQ7FXI4Y5WGXEOBISVZDX52","json":"https://pith.science/pith/Q2IDQ7FXI4Y5WGXEOBISVZDX52.json","graph_json":"https://pith.science/api/pith-number/Q2IDQ7FXI4Y5WGXEOBISVZDX52/graph.json","events_json":"https://pith.science/api/pith-number/Q2IDQ7FXI4Y5WGXEOBISVZDX52/events.json","paper":"https://pith.science/paper/Q2IDQ7FX"},"agent_actions":{"view_html":"https://pith.science/pith/Q2IDQ7FXI4Y5WGXEOBISVZDX52","download_json":"https://pith.science/pith/Q2IDQ7FXI4Y5WGXEOBISVZDX52.json","view_paper":"https://pith.science/paper/Q2IDQ7FX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.16169&json=true","fetch_graph":"https://pith.science/api/pith-number/Q2IDQ7FXI4Y5WGXEOBISVZDX52/graph.json","fetch_events":"https://pith.science/api/pith-number/Q2IDQ7FXI4Y5WGXEOBISVZDX52/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/Q2IDQ7FXI4Y5WGXEOBISVZDX52/action/timestamp_anchor","attest_storage":"https://pith.science/pith/Q2IDQ7FXI4Y5WGXEOBISVZDX52/action/storage_attestation","attest_author":"https://pith.science/pith/Q2IDQ7FXI4Y5WGXEOBISVZDX52/action/author_attestation","sign_citation":"https://pith.science/pith/Q2IDQ7FXI4Y5WGXEOBISVZDX52/action/citation_signature","submit_replication":"https://pith.science/pith/Q2IDQ7FXI4Y5WGXEOBISVZDX52/action/replication_record"}},"created_at":"2026-07-20T02:19:37.361257+00:00","updated_at":"2026-07-20T02:19:37.361257+00:00"}