{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:GOTXSE5PUKK5BCBQJAR5FMR56W","short_pith_number":"pith:GOTXSE5P","schema_version":"1.0","canonical_sha256":"33a77913afa295d088304823d2b23df593e7e5c82e3d9c14ca3bb37cb753e565","source":{"kind":"arxiv","id":"2607.20512","version":1},"attestation_state":"computed","paper":{"title":"The Active Ingredient in Muon's Grokking","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Yufeng Wang","submitted_at":"2026-07-06T22:35:57Z","abstract_excerpt":"The Muon optimizer reaches the grokking threshold on modular arithmetic faster than AdamW. Prior work attributes this to \"spectral-norm constraints plus orthogonalized momentum\" but does not isolate which mechanism matters. To better understand Moun's behavior, we run multi-seed and multi-learning-rate sweeps to decompose and stress-test the effect. First, an ablation shows the speedup comes from orthogonalization (the Newton-Schulz iteration): orthogonalize-only matches full Muon, whereas spectral-only is no faster than AdamW and is unreliable, and this verdict holds across learning rates. Se"},"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":"2607.20512","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-07-06T22:35:57Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"f0c89356cd3c875bb7934638c42d5a216592519c50d3418968515c77ef229c79","abstract_canon_sha256":"2296a0c490be950a41c961eac5cf8b10fbfacc290dd9ef6fb8a1e69387a6014d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-24T00:23:19.730099Z","signature_b64":"9cLtdnZ0b0bwVjeex4YGS6oaIPdofDwc4AXzRmv9KZZZvwkb+WA675obWZitNPLnY3ML73PZmElp2Ijy6UpsDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"33a77913afa295d088304823d2b23df593e7e5c82e3d9c14ca3bb37cb753e565","last_reissued_at":"2026-07-24T00:23:19.729243Z","signature_status":"signed_v1","first_computed_at":"2026-07-24T00:23:19.729243Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The Active Ingredient in Muon's Grokking","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Yufeng Wang","submitted_at":"2026-07-06T22:35:57Z","abstract_excerpt":"The Muon optimizer reaches the grokking threshold on modular arithmetic faster than AdamW. Prior work attributes this to \"spectral-norm constraints plus orthogonalized momentum\" but does not isolate which mechanism matters. To better understand Moun's behavior, we run multi-seed and multi-learning-rate sweeps to decompose and stress-test the effect. First, an ablation shows the speedup comes from orthogonalization (the Newton-Schulz iteration): orthogonalize-only matches full Muon, whereas spectral-only is no faster than AdamW and is unreliable, and this verdict holds across learning rates. Se"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.20512","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/2607.20512/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":"2607.20512","created_at":"2026-07-24T00:23:19.729693+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.20512v1","created_at":"2026-07-24T00:23:19.729693+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.20512","created_at":"2026-07-24T00:23:19.729693+00:00"},{"alias_kind":"pith_short_12","alias_value":"GOTXSE5PUKK5","created_at":"2026-07-24T00:23:19.729693+00:00"},{"alias_kind":"pith_short_16","alias_value":"GOTXSE5PUKK5BCBQ","created_at":"2026-07-24T00:23:19.729693+00:00"},{"alias_kind":"pith_short_8","alias_value":"GOTXSE5P","created_at":"2026-07-24T00:23:19.729693+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2608.07436","citing_title":"Post-Grokking Collapse at the Representation-Readout Interface in Muon-Trained Transformers","ref_index":17,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GOTXSE5PUKK5BCBQJAR5FMR56W","json":"https://pith.science/pith/GOTXSE5PUKK5BCBQJAR5FMR56W.json","graph_json":"https://pith.science/api/pith-number/GOTXSE5PUKK5BCBQJAR5FMR56W/graph.json","events_json":"https://pith.science/api/pith-number/GOTXSE5PUKK5BCBQJAR5FMR56W/events.json","paper":"https://pith.science/paper/GOTXSE5P"},"agent_actions":{"view_html":"https://pith.science/pith/GOTXSE5PUKK5BCBQJAR5FMR56W","download_json":"https://pith.science/pith/GOTXSE5PUKK5BCBQJAR5FMR56W.json","view_paper":"https://pith.science/paper/GOTXSE5P","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.20512&json=true","fetch_graph":"https://pith.science/api/pith-number/GOTXSE5PUKK5BCBQJAR5FMR56W/graph.json","fetch_events":"https://pith.science/api/pith-number/GOTXSE5PUKK5BCBQJAR5FMR56W/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GOTXSE5PUKK5BCBQJAR5FMR56W/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GOTXSE5PUKK5BCBQJAR5FMR56W/action/storage_attestation","attest_author":"https://pith.science/pith/GOTXSE5PUKK5BCBQJAR5FMR56W/action/author_attestation","sign_citation":"https://pith.science/pith/GOTXSE5PUKK5BCBQJAR5FMR56W/action/citation_signature","submit_replication":"https://pith.science/pith/GOTXSE5PUKK5BCBQJAR5FMR56W/action/replication_record"}},"created_at":"2026-07-24T00:23:19.729693+00:00","updated_at":"2026-07-24T00:23:19.729693+00:00"}