{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:HZLVOKH7EAK254VOF576YYGXW7","short_pith_number":"pith:HZLVOKH7","schema_version":"1.0","canonical_sha256":"3e575728ff2015aef2ae2f7fec60d7b7ee979e05424cc372baf19e2173d4b17d","source":{"kind":"arxiv","id":"2604.18170","version":2},"attestation_state":"computed","paper":{"title":"Copy-as-Decode: Grammar-Constrained Parallel Prefill for LLM Editing","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Copy-as-Decode recasts LLM editing as grammar-constrained decoding that copies input spans via single-step parallel prefill instead of full autoregressive regeneration.","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Ziyang Liu","submitted_at":"2026-04-20T12:29:53Z","abstract_excerpt":"LLMs edit text and code by autoregressively regenerating the full output, even when most tokens appear verbatim in the input. We study Copy-as-Decode, a decoding-layer mechanism that recasts edit generation as structured decoding over a two-primitive grammar: <copy lines=\"i-j\"/> references an input line range, <gen>...</gen> emits new content. A token-level FSM guarantees syntactic validity, and a serving-layer primitive updates the KV cache for each copy span via a single parallel-prefill forward rather than $N$ autoregressive steps -- sharing the parallel-forward kernel of speculative decodi"},"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":"2604.18170","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-04-20T12:29:53Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"9b4be4b275a6cf057ef9a4ec5952bbc2cc1c7a90023db06737844ca715b4c59c","abstract_canon_sha256":"17efdecb043b41135baa5d878e21bba3cc2c4eb3506bb483e6ebee44fb024037"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-26T01:03:30.636267Z","signature_b64":"NzVMTCtXpSJUuBIloJG29e4I4hFNXQZXPHgnO6cVKkqgcG06RP2lIC4YfJ4buAe1+oA6MFrhWd8/yR8q195MAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3e575728ff2015aef2ae2f7fec60d7b7ee979e05424cc372baf19e2173d4b17d","last_reissued_at":"2026-05-26T01:03:30.635377Z","signature_status":"signed_v1","first_computed_at":"2026-05-26T01:03:30.635377Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Copy-as-Decode: Grammar-Constrained Parallel Prefill for LLM Editing","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Copy-as-Decode recasts LLM editing as grammar-constrained decoding that copies input spans via single-step parallel prefill instead of full autoregressive regeneration.","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Ziyang Liu","submitted_at":"2026-04-20T12:29:53Z","abstract_excerpt":"LLMs edit text and code by autoregressively regenerating the full output, even when most tokens appear verbatim in the input. We study Copy-as-Decode, a decoding-layer mechanism that recasts edit generation as structured decoding over a two-primitive grammar: <copy lines=\"i-j\"/> references an input line range, <gen>...</gen> emits new content. A token-level FSM guarantees syntactic validity, and a serving-layer primitive updates the KV cache for each copy span via a single parallel-prefill forward rather than $N$ autoregressive steps -- sharing the parallel-forward kernel of speculative decodi"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Composed with the empirical kernel over each corpus's span histogram this yields a closed-form wall-clock bound of 29.0× / 3.4× / 4.2× (13.0× pooled). A token-level extension reaches 91–99% coverage with 4.5×–6.5× floors. Oracle programs round-trip through the deterministic resolver on all 482 cases.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That accurate copy spans (line ranges or token ranges) can be selected at inference time; the mechanism itself is shown to be lossless under perfect spans, but the perturbation study demonstrates that off-by-one noise drops pooled EM from 100% to 15.48%.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Copy-as-Decode recasts LLM editing as grammar-constrained decoding over copy and generate primitives, delivering closed-form upper-bound speedups of 13x pooled on editing benchmarks via parallel prefill without any training.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Copy-as-Decode recasts LLM editing as grammar-constrained decoding that copies input spans via single-step parallel prefill instead of full autoregressive regeneration.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"73aa632e692e94de4f206ab5879156f2ad6a9bc36d86ac04c139bd8719379209"},"source":{"id":"2604.18170","kind":"arxiv","version":2},"verdict":{"id":"fa474098-3e7f-4046-ba9c-550153645f5a","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-10T04:39:45.471296Z","strongest_claim":"Composed with the empirical kernel over each corpus's span histogram this yields a closed-form wall-clock bound of 29.0× / 3.4× / 4.2× (13.0× pooled). A token-level extension reaches 91–99% coverage with 4.5×–6.5× floors. Oracle programs round-trip through the deterministic resolver on all 482 cases.","one_line_summary":"Copy-as-Decode recasts LLM editing as grammar-constrained decoding over copy and generate primitives, delivering closed-form upper-bound speedups of 13x pooled on editing benchmarks via parallel prefill without any training.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That accurate copy spans (line ranges or token ranges) can be selected at inference time; the mechanism itself is shown to be lossless under perfect spans, but the perturbation study demonstrates that off-by-one noise drops pooled EM from 100% to 15.48%.","pith_extraction_headline":"Copy-as-Decode recasts LLM editing as grammar-constrained decoding that copies input spans via single-step parallel prefill instead of full autoregressive regeneration."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.18170/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"doi_compliance","ran_at":"2026-05-20T04:23:33.036863Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"3cb60e181265e535f4938e09e63f59a14ae664f672bd51866968784e56ee9621"},"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":"2604.18170","created_at":"2026-05-26T01:03:30.635504+00:00"},{"alias_kind":"arxiv_version","alias_value":"2604.18170v2","created_at":"2026-05-26T01:03:30.635504+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.18170","created_at":"2026-05-26T01:03:30.635504+00:00"},{"alias_kind":"pith_short_12","alias_value":"HZLVOKH7EAK2","created_at":"2026-05-26T01:03:30.635504+00:00"},{"alias_kind":"pith_short_16","alias_value":"HZLVOKH7EAK254VO","created_at":"2026-05-26T01:03:30.635504+00:00"},{"alias_kind":"pith_short_8","alias_value":"HZLVOKH7","created_at":"2026-05-26T01:03:30.635504+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/HZLVOKH7EAK254VOF576YYGXW7","json":"https://pith.science/pith/HZLVOKH7EAK254VOF576YYGXW7.json","graph_json":"https://pith.science/api/pith-number/HZLVOKH7EAK254VOF576YYGXW7/graph.json","events_json":"https://pith.science/api/pith-number/HZLVOKH7EAK254VOF576YYGXW7/events.json","paper":"https://pith.science/paper/HZLVOKH7"},"agent_actions":{"view_html":"https://pith.science/pith/HZLVOKH7EAK254VOF576YYGXW7","download_json":"https://pith.science/pith/HZLVOKH7EAK254VOF576YYGXW7.json","view_paper":"https://pith.science/paper/HZLVOKH7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2604.18170&json=true","fetch_graph":"https://pith.science/api/pith-number/HZLVOKH7EAK254VOF576YYGXW7/graph.json","fetch_events":"https://pith.science/api/pith-number/HZLVOKH7EAK254VOF576YYGXW7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HZLVOKH7EAK254VOF576YYGXW7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HZLVOKH7EAK254VOF576YYGXW7/action/storage_attestation","attest_author":"https://pith.science/pith/HZLVOKH7EAK254VOF576YYGXW7/action/author_attestation","sign_citation":"https://pith.science/pith/HZLVOKH7EAK254VOF576YYGXW7/action/citation_signature","submit_replication":"https://pith.science/pith/HZLVOKH7EAK254VOF576YYGXW7/action/replication_record"}},"created_at":"2026-05-26T01:03:30.635504+00:00","updated_at":"2026-05-26T01:03:30.635504+00:00"}