{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:WIFF52GNLWDDXSAYSMZRIPXPEJ","short_pith_number":"pith:WIFF52GN","schema_version":"1.0","canonical_sha256":"b20a5ee8cd5d863bc8189333143eef2248b176ec03e3a42a5ecaefb261430364","source":{"kind":"arxiv","id":"2406.19371","version":2},"attestation_state":"computed","paper":{"title":"Suri: Multi-constraint Instruction Following for Long-form Text Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chau Minh Pham, Mohit Iyyer, Simeng Sun","submitted_at":"2024-06-27T17:50:35Z","abstract_excerpt":"Existing research on instruction following largely focuses on tasks with simple instructions and short responses. In this work, we explore multi-constraint instruction following for generating long-form text. We create Suri, a dataset with 20K human-written long-form texts paired with LLM-generated backtranslated instructions that contain multiple complex constraints. Because of prohibitive challenges associated with collecting human preference judgments on long-form texts, preference-tuning algorithms such as DPO are infeasible in our setting; thus, we propose Instructional ORPO (I-ORPO), an "},"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":"2406.19371","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-27T17:50:35Z","cross_cats_sorted":[],"title_canon_sha256":"ac7ac046a3b74d8e70c5548cbe2dbf545484497d22aa90806477ee86f65ab87b","abstract_canon_sha256":"19c9fd0886d14c88a7e4965e11a0847a8bc7a273d62bb1a776358fead4028554"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:14:35.107575Z","signature_b64":"tufgNUlSQdSzrLB25HF7REz2U7NunMhnx/n/p8LiLxnT55/14U42rYlh9tJNYB1Etqf8dBzFPZaFjJw+KYwaAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b20a5ee8cd5d863bc8189333143eef2248b176ec03e3a42a5ecaefb261430364","last_reissued_at":"2026-07-05T09:14:35.107140Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:14:35.107140Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Suri: Multi-constraint Instruction Following for Long-form Text Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chau Minh Pham, Mohit Iyyer, Simeng Sun","submitted_at":"2024-06-27T17:50:35Z","abstract_excerpt":"Existing research on instruction following largely focuses on tasks with simple instructions and short responses. In this work, we explore multi-constraint instruction following for generating long-form text. We create Suri, a dataset with 20K human-written long-form texts paired with LLM-generated backtranslated instructions that contain multiple complex constraints. Because of prohibitive challenges associated with collecting human preference judgments on long-form texts, preference-tuning algorithms such as DPO are infeasible in our setting; thus, we propose Instructional ORPO (I-ORPO), an "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.19371","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/2406.19371/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":"2406.19371","created_at":"2026-07-05T09:14:35.107201+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.19371v2","created_at":"2026-07-05T09:14:35.107201+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.19371","created_at":"2026-07-05T09:14:35.107201+00:00"},{"alias_kind":"pith_short_12","alias_value":"WIFF52GNLWDD","created_at":"2026-07-05T09:14:35.107201+00:00"},{"alias_kind":"pith_short_16","alias_value":"WIFF52GNLWDDXSAY","created_at":"2026-07-05T09:14:35.107201+00:00"},{"alias_kind":"pith_short_8","alias_value":"WIFF52GN","created_at":"2026-07-05T09:14:35.107201+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2506.18841","citing_title":"LongWriter-Zero: Mastering Ultra-Long Text Generation via Reinforcement Learning","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2511.15408","citing_title":"Chinese Short-Form Creative Content Generation via Explanation-Oriented Multi-Objective Optimization","ref_index":56,"is_internal_anchor":false},{"citing_arxiv_id":"2604.27453","citing_title":"From Coarse to Fine: Benchmarking and Reward Modeling for Writing-Centric Generation Tasks","ref_index":2,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/WIFF52GNLWDDXSAYSMZRIPXPEJ","json":"https://pith.science/pith/WIFF52GNLWDDXSAYSMZRIPXPEJ.json","graph_json":"https://pith.science/api/pith-number/WIFF52GNLWDDXSAYSMZRIPXPEJ/graph.json","events_json":"https://pith.science/api/pith-number/WIFF52GNLWDDXSAYSMZRIPXPEJ/events.json","paper":"https://pith.science/paper/WIFF52GN"},"agent_actions":{"view_html":"https://pith.science/pith/WIFF52GNLWDDXSAYSMZRIPXPEJ","download_json":"https://pith.science/pith/WIFF52GNLWDDXSAYSMZRIPXPEJ.json","view_paper":"https://pith.science/paper/WIFF52GN","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.19371&json=true","fetch_graph":"https://pith.science/api/pith-number/WIFF52GNLWDDXSAYSMZRIPXPEJ/graph.json","fetch_events":"https://pith.science/api/pith-number/WIFF52GNLWDDXSAYSMZRIPXPEJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WIFF52GNLWDDXSAYSMZRIPXPEJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WIFF52GNLWDDXSAYSMZRIPXPEJ/action/storage_attestation","attest_author":"https://pith.science/pith/WIFF52GNLWDDXSAYSMZRIPXPEJ/action/author_attestation","sign_citation":"https://pith.science/pith/WIFF52GNLWDDXSAYSMZRIPXPEJ/action/citation_signature","submit_replication":"https://pith.science/pith/WIFF52GNLWDDXSAYSMZRIPXPEJ/action/replication_record"}},"created_at":"2026-07-05T09:14:35.107201+00:00","updated_at":"2026-07-05T09:14:35.107201+00:00"}