{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:FNXAMCVAVWLDU5WUAKBZ3QVWIQ","short_pith_number":"pith:FNXAMCVA","schema_version":"1.0","canonical_sha256":"2b6e060aa0ad963a76d402839dc2b6441ed8a4a0fef854487ade17ee4788d25d","source":{"kind":"arxiv","id":"2605.19282","version":1},"attestation_state":"computed","paper":{"title":"Rethinking Muon Beyond Pretraining: Spectral Failures and High-Pass Remedies for VLA and RLVR","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Chongyu Fan, Gaowen Liu, Mingyi Hong, Ramana Rao Kompella, Sijia Liu","submitted_at":"2026-05-19T03:00:26Z","abstract_excerpt":"Muon is a matrix-aware optimizer that leverages Newton-Schulz (NS) iterations to enforce spectral gradient orthogonalization by driving all singular values of the momentum matrix toward 1. While this uniform spectral whitening enhances exploration and outperforms AdamW in LLM pretraining, we show it could lead to fundamental limitations beyond pretraining in two regimes: (i) cross-modality vision-language-action (VLA) training, where inherently low-rank action-module gradients cause amplification of noisy tail directions, and (ii) reinforcement learning with verifiable rewards (RLVR), where lo"},"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":"2605.19282","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-19T03:00:26Z","cross_cats_sorted":[],"title_canon_sha256":"40674b61336762d654e7102176a145bef729187da72f2473630c2a928fbb6102","abstract_canon_sha256":"d5e8836c779e9b16878167adf998b8e5a966780c6b5affacab5827cd7194d9c7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T01:05:37.520205Z","signature_b64":"rJnrdbmei6yYmw20rfHkaquHqGovuDPNKvn+YSLAKt1P/odnKUivgDDTolBOZS+kvRiiV6xTxQfD0v8FXM4NDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2b6e060aa0ad963a76d402839dc2b6441ed8a4a0fef854487ade17ee4788d25d","last_reissued_at":"2026-05-20T01:05:37.519410Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T01:05:37.519410Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Rethinking Muon Beyond Pretraining: Spectral Failures and High-Pass Remedies for VLA and RLVR","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Chongyu Fan, Gaowen Liu, Mingyi Hong, Ramana Rao Kompella, Sijia Liu","submitted_at":"2026-05-19T03:00:26Z","abstract_excerpt":"Muon is a matrix-aware optimizer that leverages Newton-Schulz (NS) iterations to enforce spectral gradient orthogonalization by driving all singular values of the momentum matrix toward 1. While this uniform spectral whitening enhances exploration and outperforms AdamW in LLM pretraining, we show it could lead to fundamental limitations beyond pretraining in two regimes: (i) cross-modality vision-language-action (VLA) training, where inherently low-rank action-module gradients cause amplification of noisy tail directions, and (ii) reinforcement learning with verifiable rewards (RLVR), where lo"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.19282","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/2605.19282/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":"2605.19282","created_at":"2026-05-20T01:05:37.519548+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.19282v1","created_at":"2026-05-20T01:05:37.519548+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.19282","created_at":"2026-05-20T01:05:37.519548+00:00"},{"alias_kind":"pith_short_12","alias_value":"FNXAMCVAVWLD","created_at":"2026-05-20T01:05:37.519548+00:00"},{"alias_kind":"pith_short_16","alias_value":"FNXAMCVAVWLDU5WU","created_at":"2026-05-20T01:05:37.519548+00:00"},{"alias_kind":"pith_short_8","alias_value":"FNXAMCVA","created_at":"2026-05-20T01:05:37.519548+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/FNXAMCVAVWLDU5WUAKBZ3QVWIQ","json":"https://pith.science/pith/FNXAMCVAVWLDU5WUAKBZ3QVWIQ.json","graph_json":"https://pith.science/api/pith-number/FNXAMCVAVWLDU5WUAKBZ3QVWIQ/graph.json","events_json":"https://pith.science/api/pith-number/FNXAMCVAVWLDU5WUAKBZ3QVWIQ/events.json","paper":"https://pith.science/paper/FNXAMCVA"},"agent_actions":{"view_html":"https://pith.science/pith/FNXAMCVAVWLDU5WUAKBZ3QVWIQ","download_json":"https://pith.science/pith/FNXAMCVAVWLDU5WUAKBZ3QVWIQ.json","view_paper":"https://pith.science/paper/FNXAMCVA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.19282&json=true","fetch_graph":"https://pith.science/api/pith-number/FNXAMCVAVWLDU5WUAKBZ3QVWIQ/graph.json","fetch_events":"https://pith.science/api/pith-number/FNXAMCVAVWLDU5WUAKBZ3QVWIQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FNXAMCVAVWLDU5WUAKBZ3QVWIQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FNXAMCVAVWLDU5WUAKBZ3QVWIQ/action/storage_attestation","attest_author":"https://pith.science/pith/FNXAMCVAVWLDU5WUAKBZ3QVWIQ/action/author_attestation","sign_citation":"https://pith.science/pith/FNXAMCVAVWLDU5WUAKBZ3QVWIQ/action/citation_signature","submit_replication":"https://pith.science/pith/FNXAMCVAVWLDU5WUAKBZ3QVWIQ/action/replication_record"}},"created_at":"2026-05-20T01:05:37.519548+00:00","updated_at":"2026-05-20T01:05:37.519548+00:00"}