{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:VIAW4PMPOHYVVJYYTWBK3JH7F5","short_pith_number":"pith:VIAW4PMP","schema_version":"1.0","canonical_sha256":"aa016e3d8f71f15aa7189d82ada4ff2f70492113faeb61831650fef2c777bd7e","source":{"kind":"arxiv","id":"2605.07066","version":2},"attestation_state":"computed","paper":{"title":"2.5-D Decomposition for LLM-Based Spatial Construction","license":"http://creativecommons.org/licenses/by/4.0/","headline":"A 2.5-D decomposition lets LLMs build structures from language instructions by planning only the horizontal plane while a deterministic executor computes vertical placements from column occupancy.","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Li-Jen Chen, Paul Whitten, Sharath Baddam","submitted_at":"2026-05-08T00:17:33Z","abstract_excerpt":"Autonomous systems that build structures from natural-language instructions need reliable spatial reasoning, yet large language models (LLMs) make systematic coordinate errors when generating three-dimensional block placements. We present a neuro-symbolic pipeline based on \\emph{2.5-D decomposition}: the LLM plans in the two-dimensional horizontal plane while a deterministic executor computes all vertical placement from column occupancy, eliminating an entire class of errors. On the Build What I Mean benchmark (160 rounds), GPT-4o-mini with this pipeline achieves 94.6\\% mean structural accurac"},"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":"2605.07066","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-08T00:17:33Z","cross_cats_sorted":[],"title_canon_sha256":"5b529320c8728de9af002b36202f427c6b3259fc19b53fdeb0e86748b126dabe","abstract_canon_sha256":"1542e9f7a33f4ca5981b4c90c26d920c3faa5a066e154bd717d7e16a69cc286a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T01:06:09.883264Z","signature_b64":"VrZ87GphDdCZMEI6WOkjWPo3jgxrA5byWigqK3Bzb5IgUY9uaEyDgvGy5QAJVRggQB9SVku8CslhjX/0yKonBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"aa016e3d8f71f15aa7189d82ada4ff2f70492113faeb61831650fef2c777bd7e","last_reissued_at":"2026-05-20T01:06:09.882679Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T01:06:09.882679Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"2.5-D Decomposition for LLM-Based Spatial Construction","license":"http://creativecommons.org/licenses/by/4.0/","headline":"A 2.5-D decomposition lets LLMs build structures from language instructions by planning only the horizontal plane while a deterministic executor computes vertical placements from column occupancy.","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Li-Jen Chen, Paul Whitten, Sharath Baddam","submitted_at":"2026-05-08T00:17:33Z","abstract_excerpt":"Autonomous systems that build structures from natural-language instructions need reliable spatial reasoning, yet large language models (LLMs) make systematic coordinate errors when generating three-dimensional block placements. We present a neuro-symbolic pipeline based on \\emph{2.5-D decomposition}: the LLM plans in the two-dimensional horizontal plane while a deterministic executor computes all vertical placement from column occupancy, eliminating an entire class of errors. On the Build What I Mean benchmark (160 rounds), GPT-4o-mini with this pipeline achieves 94.6\\% mean structural accurac"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"On the Build What I Mean benchmark (160 rounds), GPT-4o-mini with this pipeline achieves 94.6% mean structural accuracy across 12 independent runs, within 3.0 percentage points of the 97.6% ceiling imposed by architect-agent errors that no builder-side improvement can address. This outperforms both GPT-4o at 90.3% and the best competing system at 76.3%.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That vertical block placements are fully and correctly determined solely by column occupancy without requiring additional spatial reasoning or handling complex inter-block dependencies beyond simple stacking.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"2.5-D decomposition lets LLMs achieve 94.6% structural accuracy on a building benchmark by handling only horizontal planning while a symbolic system manages vertical placements from occupancy.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A 2.5-D decomposition lets LLMs build structures from language instructions by planning only the horizontal plane while a deterministic executor computes vertical placements from column occupancy.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"ce6d46e50058680625448091d3675ec734afa4175bc4a8fc1950f888b6b93851"},"source":{"id":"2605.07066","kind":"arxiv","version":2},"verdict":{"id":"0b1a9de1-ebbc-4e74-9271-2e98c156cbdb","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-11T01:31:26.926216Z","strongest_claim":"On the Build What I Mean benchmark (160 rounds), GPT-4o-mini with this pipeline achieves 94.6% mean structural accuracy across 12 independent runs, within 3.0 percentage points of the 97.6% ceiling imposed by architect-agent errors that no builder-side improvement can address. This outperforms both GPT-4o at 90.3% and the best competing system at 76.3%.","one_line_summary":"2.5-D decomposition lets LLMs achieve 94.6% structural accuracy on a building benchmark by handling only horizontal planning while a symbolic system manages vertical placements from occupancy.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That vertical block placements are fully and correctly determined solely by column occupancy without requiring additional spatial reasoning or handling complex inter-block dependencies beyond simple stacking.","pith_extraction_headline":"A 2.5-D decomposition lets LLMs build structures from language instructions by planning only the horizontal plane while a deterministic executor computes vertical placements from column occupancy."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.07066/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"doi_title_agreement","ran_at":"2026-05-19T17:31:18.715353Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T12:09:30.060770Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"9c04d00e8a73e396ca98832b3ba26f9a3eedfe47f9447eb17ff63c5acce6153a"},"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":"2605.07066","created_at":"2026-05-20T01:06:09.882760+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.07066v2","created_at":"2026-05-20T01:06:09.882760+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.07066","created_at":"2026-05-20T01:06:09.882760+00:00"},{"alias_kind":"pith_short_12","alias_value":"VIAW4PMPOHYV","created_at":"2026-05-20T01:06:09.882760+00:00"},{"alias_kind":"pith_short_16","alias_value":"VIAW4PMPOHYVVJYY","created_at":"2026-05-20T01:06:09.882760+00:00"},{"alias_kind":"pith_short_8","alias_value":"VIAW4PMP","created_at":"2026-05-20T01:06:09.882760+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/VIAW4PMPOHYVVJYYTWBK3JH7F5","json":"https://pith.science/pith/VIAW4PMPOHYVVJYYTWBK3JH7F5.json","graph_json":"https://pith.science/api/pith-number/VIAW4PMPOHYVVJYYTWBK3JH7F5/graph.json","events_json":"https://pith.science/api/pith-number/VIAW4PMPOHYVVJYYTWBK3JH7F5/events.json","paper":"https://pith.science/paper/VIAW4PMP"},"agent_actions":{"view_html":"https://pith.science/pith/VIAW4PMPOHYVVJYYTWBK3JH7F5","download_json":"https://pith.science/pith/VIAW4PMPOHYVVJYYTWBK3JH7F5.json","view_paper":"https://pith.science/paper/VIAW4PMP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.07066&json=true","fetch_graph":"https://pith.science/api/pith-number/VIAW4PMPOHYVVJYYTWBK3JH7F5/graph.json","fetch_events":"https://pith.science/api/pith-number/VIAW4PMPOHYVVJYYTWBK3JH7F5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VIAW4PMPOHYVVJYYTWBK3JH7F5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VIAW4PMPOHYVVJYYTWBK3JH7F5/action/storage_attestation","attest_author":"https://pith.science/pith/VIAW4PMPOHYVVJYYTWBK3JH7F5/action/author_attestation","sign_citation":"https://pith.science/pith/VIAW4PMPOHYVVJYYTWBK3JH7F5/action/citation_signature","submit_replication":"https://pith.science/pith/VIAW4PMPOHYVVJYYTWBK3JH7F5/action/replication_record"}},"created_at":"2026-05-20T01:06:09.882760+00:00","updated_at":"2026-05-20T01:06:09.882760+00:00"}