{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:BZLTA3UVYKXVBBCWSXHDHLSAR6","short_pith_number":"pith:BZLTA3UV","schema_version":"1.0","canonical_sha256":"0e57306e95c2af50845695ce33ae408fbd2c0831d0e1a73be115023c4a3d00d9","source":{"kind":"arxiv","id":"2505.07067","version":1},"attestation_state":"computed","paper":{"title":"Learning curves theory for hierarchically compositional data with power-law distributed features","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cond-mat.dis-nn","cs.LG"],"primary_cat":"stat.ML","authors_text":"Francesco Cagnetta, Hyunmo Kang, Matthieu Wyart","submitted_at":"2025-05-11T17:38:40Z","abstract_excerpt":"Recent theories suggest that Neural Scaling Laws arise whenever the task is linearly decomposed into power-law distributed units. Alternatively, scaling laws also emerge when data exhibit a hierarchically compositional structure, as is thought to occur in language and images. To unify these views, we consider classification and next-token prediction tasks based on probabilistic context-free grammars -- probabilistic models that generate data via a hierarchy of production rules. For classification, we show that having power-law distributed production rules results in a power-law learning curve "},"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":"2505.07067","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"stat.ML","submitted_at":"2025-05-11T17:38:40Z","cross_cats_sorted":["cond-mat.dis-nn","cs.LG"],"title_canon_sha256":"229551290ac4d2b9cce6bd5e76b30f20fa563cbb8455df94438a22aad4f8c8db","abstract_canon_sha256":"98e845813f0d39c10cec34d8e902d034f653b4572c2bbf8955dbd68159eab4a8"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:01:42.224350Z","signature_b64":"U0itwy+PqT4z5AtO2TJ1WQ1r0nJCNU3emKfAKiI4njNI12VW2KvDz+IX6EnjE7xjOFZ/bVD8oRmcHYiaPrz5Dg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0e57306e95c2af50845695ce33ae408fbd2c0831d0e1a73be115023c4a3d00d9","last_reissued_at":"2026-07-05T11:01:42.223848Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:01:42.223848Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning curves theory for hierarchically compositional data with power-law distributed features","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cond-mat.dis-nn","cs.LG"],"primary_cat":"stat.ML","authors_text":"Francesco Cagnetta, Hyunmo Kang, Matthieu Wyart","submitted_at":"2025-05-11T17:38:40Z","abstract_excerpt":"Recent theories suggest that Neural Scaling Laws arise whenever the task is linearly decomposed into power-law distributed units. Alternatively, scaling laws also emerge when data exhibit a hierarchically compositional structure, as is thought to occur in language and images. To unify these views, we consider classification and next-token prediction tasks based on probabilistic context-free grammars -- probabilistic models that generate data via a hierarchy of production rules. For classification, we show that having power-law distributed production rules results in a power-law learning curve "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.07067","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/2505.07067/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":"2505.07067","created_at":"2026-07-05T11:01:42.223916+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.07067v1","created_at":"2026-07-05T11:01:42.223916+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.07067","created_at":"2026-07-05T11:01:42.223916+00:00"},{"alias_kind":"pith_short_12","alias_value":"BZLTA3UVYKXV","created_at":"2026-07-05T11:01:42.223916+00:00"},{"alias_kind":"pith_short_16","alias_value":"BZLTA3UVYKXVBBCW","created_at":"2026-07-05T11:01:42.223916+00:00"},{"alias_kind":"pith_short_8","alias_value":"BZLTA3UV","created_at":"2026-07-05T11:01:42.223916+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.20347","citing_title":"Critical Percolation as a Synthetic Data Model for Interpretability","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10395","citing_title":"Sharp feature-learning transitions and Bayes-optimal neural scaling laws in extensive-width networks","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2604.21691","citing_title":"There Will Be a Scientific Theory of Deep Learning","ref_index":249,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/BZLTA3UVYKXVBBCWSXHDHLSAR6","json":"https://pith.science/pith/BZLTA3UVYKXVBBCWSXHDHLSAR6.json","graph_json":"https://pith.science/api/pith-number/BZLTA3UVYKXVBBCWSXHDHLSAR6/graph.json","events_json":"https://pith.science/api/pith-number/BZLTA3UVYKXVBBCWSXHDHLSAR6/events.json","paper":"https://pith.science/paper/BZLTA3UV"},"agent_actions":{"view_html":"https://pith.science/pith/BZLTA3UVYKXVBBCWSXHDHLSAR6","download_json":"https://pith.science/pith/BZLTA3UVYKXVBBCWSXHDHLSAR6.json","view_paper":"https://pith.science/paper/BZLTA3UV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.07067&json=true","fetch_graph":"https://pith.science/api/pith-number/BZLTA3UVYKXVBBCWSXHDHLSAR6/graph.json","fetch_events":"https://pith.science/api/pith-number/BZLTA3UVYKXVBBCWSXHDHLSAR6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BZLTA3UVYKXVBBCWSXHDHLSAR6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BZLTA3UVYKXVBBCWSXHDHLSAR6/action/storage_attestation","attest_author":"https://pith.science/pith/BZLTA3UVYKXVBBCWSXHDHLSAR6/action/author_attestation","sign_citation":"https://pith.science/pith/BZLTA3UVYKXVBBCWSXHDHLSAR6/action/citation_signature","submit_replication":"https://pith.science/pith/BZLTA3UVYKXVBBCWSXHDHLSAR6/action/replication_record"}},"created_at":"2026-07-05T11:01:42.223916+00:00","updated_at":"2026-07-05T11:01:42.223916+00:00"}