{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:O4RKBOYVF63QVGAEAAV3YDQBTD","short_pith_number":"pith:O4RKBOYV","schema_version":"1.0","canonical_sha256":"7722a0bb152fb70a9804002bbc0e0198d9bf18694373eaedd5a3c3f5604a1e57","source":{"kind":"arxiv","id":"2505.14425","version":2},"attestation_state":"computed","paper":{"title":"From Templates to Natural Language: Generalization Challenges in Instruction-Tuned LLMs for Spatial Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chalamalasetti Kranti, David Schlangen, Sherzod Hakimov","submitted_at":"2025-05-20T14:33:29Z","abstract_excerpt":"Instruction-tuned large language models (LLMs) have shown strong performance on a variety of tasks; however, generalizing from synthetic to human-authored instructions in grounded environments remains a challenge for them. In this work, we study generalization challenges in spatial grounding tasks where models interpret and translate instructions for building object arrangements on a $2.5$D grid. We fine-tune LLMs using only synthetic instructions and evaluate their performance on a benchmark dataset containing both synthetic and human-written instructions. Our results reveal that while models"},"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":"2505.14425","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-05-20T14:33:29Z","cross_cats_sorted":[],"title_canon_sha256":"418766be6bab8234a73597f942d7fc40a25cdc969a32ca0dcbfea26499d55641","abstract_canon_sha256":"7f5b275841cc6a3c3b3df198c585d7e095f6ebcd72120bcdd518d4d4462233f0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:55:19.497014Z","signature_b64":"0sT2CeJ95giih7eAjVfEmxqjfFjNenXFXZgGifjuW1zo285sT8ZNwtfLtDGcIEZVij3pw/LdNABA9/pcZ6idBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7722a0bb152fb70a9804002bbc0e0198d9bf18694373eaedd5a3c3f5604a1e57","last_reissued_at":"2026-07-05T11:55:19.496554Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:55:19.496554Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"From Templates to Natural Language: Generalization Challenges in Instruction-Tuned LLMs for Spatial Reasoning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chalamalasetti Kranti, David Schlangen, Sherzod Hakimov","submitted_at":"2025-05-20T14:33:29Z","abstract_excerpt":"Instruction-tuned large language models (LLMs) have shown strong performance on a variety of tasks; however, generalizing from synthetic to human-authored instructions in grounded environments remains a challenge for them. In this work, we study generalization challenges in spatial grounding tasks where models interpret and translate instructions for building object arrangements on a $2.5$D grid. We fine-tune LLMs using only synthetic instructions and evaluate their performance on a benchmark dataset containing both synthetic and human-written instructions. Our results reveal that while models"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.14425","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/2505.14425/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":"2505.14425","created_at":"2026-07-05T11:55:19.496605+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.14425v2","created_at":"2026-07-05T11:55:19.496605+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.14425","created_at":"2026-07-05T11:55:19.496605+00:00"},{"alias_kind":"pith_short_12","alias_value":"O4RKBOYVF63Q","created_at":"2026-07-05T11:55:19.496605+00:00"},{"alias_kind":"pith_short_16","alias_value":"O4RKBOYVF63QVGAE","created_at":"2026-07-05T11:55:19.496605+00:00"},{"alias_kind":"pith_short_8","alias_value":"O4RKBOYV","created_at":"2026-07-05T11:55:19.496605+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/O4RKBOYVF63QVGAEAAV3YDQBTD","json":"https://pith.science/pith/O4RKBOYVF63QVGAEAAV3YDQBTD.json","graph_json":"https://pith.science/api/pith-number/O4RKBOYVF63QVGAEAAV3YDQBTD/graph.json","events_json":"https://pith.science/api/pith-number/O4RKBOYVF63QVGAEAAV3YDQBTD/events.json","paper":"https://pith.science/paper/O4RKBOYV"},"agent_actions":{"view_html":"https://pith.science/pith/O4RKBOYVF63QVGAEAAV3YDQBTD","download_json":"https://pith.science/pith/O4RKBOYVF63QVGAEAAV3YDQBTD.json","view_paper":"https://pith.science/paper/O4RKBOYV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.14425&json=true","fetch_graph":"https://pith.science/api/pith-number/O4RKBOYVF63QVGAEAAV3YDQBTD/graph.json","fetch_events":"https://pith.science/api/pith-number/O4RKBOYVF63QVGAEAAV3YDQBTD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/O4RKBOYVF63QVGAEAAV3YDQBTD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/O4RKBOYVF63QVGAEAAV3YDQBTD/action/storage_attestation","attest_author":"https://pith.science/pith/O4RKBOYVF63QVGAEAAV3YDQBTD/action/author_attestation","sign_citation":"https://pith.science/pith/O4RKBOYVF63QVGAEAAV3YDQBTD/action/citation_signature","submit_replication":"https://pith.science/pith/O4RKBOYVF63QVGAEAAV3YDQBTD/action/replication_record"}},"created_at":"2026-07-05T11:55:19.496605+00:00","updated_at":"2026-07-05T11:55:19.496605+00:00"}