{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:KYZN76JHHR7P4UDIGRJXCARWFK","short_pith_number":"pith:KYZN76JH","schema_version":"1.0","canonical_sha256":"5632dff9273c7efe506834537102362ab3d582df68a859f2bf3b6039fd45ef29","source":{"kind":"arxiv","id":"2605.25966","version":1},"attestation_state":"computed","paper":{"title":"Mapping the Schedule x Bit-Width Boundary in Sub-100M Quantisation-Aware Training","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Christian Brandt Thomassen","submitted_at":"2026-05-25T15:42:34Z","abstract_excerpt":"We test whether the optimal learning-rate schedule depends on bit-width during from-initialisation quantisation-aware training (QAT) for sub-100M decoder language models. A 720-run factorial grid (Phase 2) over bit-width x warmdown fraction x LR magnitude x model size x seed (FP16/INT8/INT6, 15M-100M, 5 seeds) finds the optimal warmdown is 33% at every (bit-width, size) cell. The primary hypothesis -- that INT6 QAT requires a different schedule than higher-precision training -- is falsified at FP16/INT8/INT6. A 625-run follow-up (Phase 5) probes the null along five axes: optimiser (AdamW), sch"},"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":"2605.25966","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-25T15:42:34Z","cross_cats_sorted":["cs.CL","stat.ML"],"title_canon_sha256":"c0ad2d491de851e0d7076273a6e29edd200b63355b75f6ab378f4fa59a24448e","abstract_canon_sha256":"8d35a5ed4bf2ca8037f3b35adc30bfcef8527a06d11372cc6ec8d0768cff36cf"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-26T02:05:20.804358Z","signature_b64":"WYa7JZ/sSD6JLIaactvJkNq8nW+/L9kOEzIBOjvR3he4fJXtzfxvd2DTNVPwpcqf8kF1Ldc32Atb84fsJ0iUAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5632dff9273c7efe506834537102362ab3d582df68a859f2bf3b6039fd45ef29","last_reissued_at":"2026-05-26T02:05:20.803759Z","signature_status":"signed_v1","first_computed_at":"2026-05-26T02:05:20.803759Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Mapping the Schedule x Bit-Width Boundary in Sub-100M Quantisation-Aware Training","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Christian Brandt Thomassen","submitted_at":"2026-05-25T15:42:34Z","abstract_excerpt":"We test whether the optimal learning-rate schedule depends on bit-width during from-initialisation quantisation-aware training (QAT) for sub-100M decoder language models. A 720-run factorial grid (Phase 2) over bit-width x warmdown fraction x LR magnitude x model size x seed (FP16/INT8/INT6, 15M-100M, 5 seeds) finds the optimal warmdown is 33% at every (bit-width, size) cell. The primary hypothesis -- that INT6 QAT requires a different schedule than higher-precision training -- is falsified at FP16/INT8/INT6. A 625-run follow-up (Phase 5) probes the null along five axes: optimiser (AdamW), sch"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.25966","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/2605.25966/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":"2605.25966","created_at":"2026-05-26T02:05:20.803856+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.25966v1","created_at":"2026-05-26T02:05:20.803856+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.25966","created_at":"2026-05-26T02:05:20.803856+00:00"},{"alias_kind":"pith_short_12","alias_value":"KYZN76JHHR7P","created_at":"2026-05-26T02:05:20.803856+00:00"},{"alias_kind":"pith_short_16","alias_value":"KYZN76JHHR7P4UDI","created_at":"2026-05-26T02:05:20.803856+00:00"},{"alias_kind":"pith_short_8","alias_value":"KYZN76JH","created_at":"2026-05-26T02:05:20.803856+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/KYZN76JHHR7P4UDIGRJXCARWFK","json":"https://pith.science/pith/KYZN76JHHR7P4UDIGRJXCARWFK.json","graph_json":"https://pith.science/api/pith-number/KYZN76JHHR7P4UDIGRJXCARWFK/graph.json","events_json":"https://pith.science/api/pith-number/KYZN76JHHR7P4UDIGRJXCARWFK/events.json","paper":"https://pith.science/paper/KYZN76JH"},"agent_actions":{"view_html":"https://pith.science/pith/KYZN76JHHR7P4UDIGRJXCARWFK","download_json":"https://pith.science/pith/KYZN76JHHR7P4UDIGRJXCARWFK.json","view_paper":"https://pith.science/paper/KYZN76JH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.25966&json=true","fetch_graph":"https://pith.science/api/pith-number/KYZN76JHHR7P4UDIGRJXCARWFK/graph.json","fetch_events":"https://pith.science/api/pith-number/KYZN76JHHR7P4UDIGRJXCARWFK/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KYZN76JHHR7P4UDIGRJXCARWFK/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KYZN76JHHR7P4UDIGRJXCARWFK/action/storage_attestation","attest_author":"https://pith.science/pith/KYZN76JHHR7P4UDIGRJXCARWFK/action/author_attestation","sign_citation":"https://pith.science/pith/KYZN76JHHR7P4UDIGRJXCARWFK/action/citation_signature","submit_replication":"https://pith.science/pith/KYZN76JHHR7P4UDIGRJXCARWFK/action/replication_record"}},"created_at":"2026-05-26T02:05:20.803856+00:00","updated_at":"2026-05-26T02:05:20.803856+00:00"}