{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:72YYAD5HHZ7EB32WBKOKIGV7MF","short_pith_number":"pith:72YYAD5H","schema_version":"1.0","canonical_sha256":"feb1800fa73e7e40ef560a9ca41abf6167db79cf5b7eb5fcd51bc0c0a67ae481","source":{"kind":"arxiv","id":"2603.00910","version":2},"attestation_state":"computed","paper":{"title":"Curvature-Weighted Capacity Allocation: A Minimum Description Length Framework for Layer-Adaptive Large Language Model Optimization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG","math.IT"],"primary_cat":"cs.IT","authors_text":"Ankur Mali, Anshuman Chhabra, Hitesh Vaidya, Theophilus Amaefuna","submitted_at":"2026-03-01T04:14:15Z","abstract_excerpt":"Layer-wise capacity in large language models is highly non-uniform: some layers contribute disproportionately to loss reduction, whereas others are nearly redundant. Existing layer-scoring methods provide sensitivity estimates but do not give a principled rule for converting those estimates into allocation or pruning decisions under a global hardware budget. We introduce a curvature-aware, MDL-inspired framework built around the layer gain $\\zeta_k^2=g_k^\\top\\widetilde H_{kk}^{-1}g_k$. This quantity equals twice the maximal decrease predicted by the regularized layer-restricted quadratic model"},"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":"2603.00910","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.IT","submitted_at":"2026-03-01T04:14:15Z","cross_cats_sorted":["cs.AI","cs.LG","math.IT"],"title_canon_sha256":"daef19f2e06740172b44f767a9786d046737f8f48aa69a70ebb7d6e4805df071","abstract_canon_sha256":"a476c9bf5769c84ede2aa2bf9d93bdf9f5e04a12e6def004266734ea8018677c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-10T00:18:44.026425Z","signature_b64":"CmaS/eIgNIx7Dqlq4gs9WKD2KOTk5V33DM9fNKJO15671dpP92pyPpw7LYkpLTqnhHxNp63NBR10AIfqG0hKDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"feb1800fa73e7e40ef560a9ca41abf6167db79cf5b7eb5fcd51bc0c0a67ae481","last_reissued_at":"2026-07-10T00:18:44.025916Z","signature_status":"signed_v1","first_computed_at":"2026-07-10T00:18:44.025916Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Curvature-Weighted Capacity Allocation: A Minimum Description Length Framework for Layer-Adaptive Large Language Model Optimization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG","math.IT"],"primary_cat":"cs.IT","authors_text":"Ankur Mali, Anshuman Chhabra, Hitesh Vaidya, Theophilus Amaefuna","submitted_at":"2026-03-01T04:14:15Z","abstract_excerpt":"Layer-wise capacity in large language models is highly non-uniform: some layers contribute disproportionately to loss reduction, whereas others are nearly redundant. Existing layer-scoring methods provide sensitivity estimates but do not give a principled rule for converting those estimates into allocation or pruning decisions under a global hardware budget. We introduce a curvature-aware, MDL-inspired framework built around the layer gain $\\zeta_k^2=g_k^\\top\\widetilde H_{kk}^{-1}g_k$. This quantity equals twice the maximal decrease predicted by the regularized layer-restricted quadratic model"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2603.00910","kind":"arxiv","version":2},"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/2603.00910/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":"2603.00910","created_at":"2026-07-10T00:18:44.025976+00:00"},{"alias_kind":"arxiv_version","alias_value":"2603.00910v2","created_at":"2026-07-10T00:18:44.025976+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2603.00910","created_at":"2026-07-10T00:18:44.025976+00:00"},{"alias_kind":"pith_short_12","alias_value":"72YYAD5HHZ7E","created_at":"2026-07-10T00:18:44.025976+00:00"},{"alias_kind":"pith_short_16","alias_value":"72YYAD5HHZ7EB32W","created_at":"2026-07-10T00:18:44.025976+00:00"},{"alias_kind":"pith_short_8","alias_value":"72YYAD5H","created_at":"2026-07-10T00:18:44.025976+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/72YYAD5HHZ7EB32WBKOKIGV7MF","json":"https://pith.science/pith/72YYAD5HHZ7EB32WBKOKIGV7MF.json","graph_json":"https://pith.science/api/pith-number/72YYAD5HHZ7EB32WBKOKIGV7MF/graph.json","events_json":"https://pith.science/api/pith-number/72YYAD5HHZ7EB32WBKOKIGV7MF/events.json","paper":"https://pith.science/paper/72YYAD5H"},"agent_actions":{"view_html":"https://pith.science/pith/72YYAD5HHZ7EB32WBKOKIGV7MF","download_json":"https://pith.science/pith/72YYAD5HHZ7EB32WBKOKIGV7MF.json","view_paper":"https://pith.science/paper/72YYAD5H","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2603.00910&json=true","fetch_graph":"https://pith.science/api/pith-number/72YYAD5HHZ7EB32WBKOKIGV7MF/graph.json","fetch_events":"https://pith.science/api/pith-number/72YYAD5HHZ7EB32WBKOKIGV7MF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/72YYAD5HHZ7EB32WBKOKIGV7MF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/72YYAD5HHZ7EB32WBKOKIGV7MF/action/storage_attestation","attest_author":"https://pith.science/pith/72YYAD5HHZ7EB32WBKOKIGV7MF/action/author_attestation","sign_citation":"https://pith.science/pith/72YYAD5HHZ7EB32WBKOKIGV7MF/action/citation_signature","submit_replication":"https://pith.science/pith/72YYAD5HHZ7EB32WBKOKIGV7MF/action/replication_record"}},"created_at":"2026-07-10T00:18:44.025976+00:00","updated_at":"2026-07-10T00:18:44.025976+00:00"}