{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:7GXYSGPBJT2RREMBUGTRSDOUSZ","short_pith_number":"pith:7GXYSGPB","schema_version":"1.0","canonical_sha256":"f9af8919e14cf5189181a1a7190dd49677b5e11b558f69ef0a07d719fd2d5084","source":{"kind":"arxiv","id":"2605.18807","version":1},"attestation_state":"computed","paper":{"title":"Block-Based Double Decoders","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Asher Labovich, Benjamin Bradley, Chaitanya Harsha, Vanessa Alexander","submitted_at":"2026-05-11T22:41:03Z","abstract_excerpt":"Encoder-decoder models offer substantial inference-time savings over decoder-only models, but their pretraining objectives suffer from sparse supervision and dynamic sequence lengths, keeping them out of practice at scale. We propose block-based double decoders, a novel transformer architecture that utilizes doubly-causal block-based attention masks to train with full loss supervision and static sequence packing, combining decoder-only training efficiency with encoder-decoder inference efficiency. In scaling law experiments, block-based double decoders strongly outperform encoder-decoders and "},"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":"2605.18807","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-11T22:41:03Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"7c8444027d34eed02eda7f6f3fd77482c70c8ede8bbb5c559daa756d8bce77b5","abstract_canon_sha256":"8652726481bcada56ca987ba5e73bda6c891d1dd62bdcce50a9a9827244895de"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:06:23.479258Z","signature_b64":"icuFg/uoEgFnFLqFVDKSutysXgJnM2PSQPkN+hR/stL5v/YYh4nMbJWjHl9rLjn8EBgOUc1Cuf4Lqk6hkX1sBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f9af8919e14cf5189181a1a7190dd49677b5e11b558f69ef0a07d719fd2d5084","last_reissued_at":"2026-05-20T00:06:23.478526Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:06:23.478526Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Block-Based Double Decoders","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Asher Labovich, Benjamin Bradley, Chaitanya Harsha, Vanessa Alexander","submitted_at":"2026-05-11T22:41:03Z","abstract_excerpt":"Encoder-decoder models offer substantial inference-time savings over decoder-only models, but their pretraining objectives suffer from sparse supervision and dynamic sequence lengths, keeping them out of practice at scale. We propose block-based double decoders, a novel transformer architecture that utilizes doubly-causal block-based attention masks to train with full loss supervision and static sequence packing, combining decoder-only training efficiency with encoder-decoder inference efficiency. In scaling law experiments, block-based double decoders strongly outperform encoder-decoders and "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.18807","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/2605.18807/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":"2605.18807","created_at":"2026-05-20T00:06:23.478662+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.18807v1","created_at":"2026-05-20T00:06:23.478662+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.18807","created_at":"2026-05-20T00:06:23.478662+00:00"},{"alias_kind":"pith_short_12","alias_value":"7GXYSGPBJT2R","created_at":"2026-05-20T00:06:23.478662+00:00"},{"alias_kind":"pith_short_16","alias_value":"7GXYSGPBJT2RREMB","created_at":"2026-05-20T00:06:23.478662+00:00"},{"alias_kind":"pith_short_8","alias_value":"7GXYSGPB","created_at":"2026-05-20T00:06:23.478662+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/7GXYSGPBJT2RREMBUGTRSDOUSZ","json":"https://pith.science/pith/7GXYSGPBJT2RREMBUGTRSDOUSZ.json","graph_json":"https://pith.science/api/pith-number/7GXYSGPBJT2RREMBUGTRSDOUSZ/graph.json","events_json":"https://pith.science/api/pith-number/7GXYSGPBJT2RREMBUGTRSDOUSZ/events.json","paper":"https://pith.science/paper/7GXYSGPB"},"agent_actions":{"view_html":"https://pith.science/pith/7GXYSGPBJT2RREMBUGTRSDOUSZ","download_json":"https://pith.science/pith/7GXYSGPBJT2RREMBUGTRSDOUSZ.json","view_paper":"https://pith.science/paper/7GXYSGPB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.18807&json=true","fetch_graph":"https://pith.science/api/pith-number/7GXYSGPBJT2RREMBUGTRSDOUSZ/graph.json","fetch_events":"https://pith.science/api/pith-number/7GXYSGPBJT2RREMBUGTRSDOUSZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7GXYSGPBJT2RREMBUGTRSDOUSZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7GXYSGPBJT2RREMBUGTRSDOUSZ/action/storage_attestation","attest_author":"https://pith.science/pith/7GXYSGPBJT2RREMBUGTRSDOUSZ/action/author_attestation","sign_citation":"https://pith.science/pith/7GXYSGPBJT2RREMBUGTRSDOUSZ/action/citation_signature","submit_replication":"https://pith.science/pith/7GXYSGPBJT2RREMBUGTRSDOUSZ/action/replication_record"}},"created_at":"2026-05-20T00:06:23.478662+00:00","updated_at":"2026-05-20T00:06:23.478662+00:00"}