{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:NZD3JL7YRQA4WSHGGYAPV6SIKX","short_pith_number":"pith:NZD3JL7Y","schema_version":"1.0","canonical_sha256":"6e47b4aff88c01cb48e63600fafa4855f7240a2993b85c053f1913d4d993b06d","source":{"kind":"arxiv","id":"2412.14656","version":1},"attestation_state":"computed","paper":{"title":"Length Controlled Generation for Black-box LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Bing Qin, Kun Zhu, Lei Huang, Tat-Seng Chua, Weihong Zhong, Wenjie Wang, Xiaocheng Feng, Yuxuan Gu","submitted_at":"2024-12-19T09:07:38Z","abstract_excerpt":"Large language models (LLMs) have demonstrated impressive instruction following capabilities, while still struggling to accurately manage the length of the generated text, which is a fundamental requirement in many real-world applications. Existing length control methods involve fine-tuning the parameters of LLMs, which is inefficient and suboptimal for practical use. In this paper, we propose a novel iterative sampling framework for text length control, integrating the Metropolis-Hastings algorithm with an importance sampling acceleration strategy. This framework efficiently and reliably regu"},"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":"2412.14656","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-12-19T09:07:38Z","cross_cats_sorted":[],"title_canon_sha256":"c05d7d58466530285b555e5c7866cc9e037bb30676fbc4e66df1b5d3b6321cd0","abstract_canon_sha256":"39862333c40a85dfd1f9aa45e3d1b30f6974fd04aff0a70ad4851bae58307bd9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:51:50.870274Z","signature_b64":"QJCyeGaVUPbHUlQMI/pZsQpA3L0TsHGxd8awudbwAuWHkFIVGezODTRu2VpnT9kP0wUdlQmQODiYbzmhhRlBCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6e47b4aff88c01cb48e63600fafa4855f7240a2993b85c053f1913d4d993b06d","last_reissued_at":"2026-07-05T09:51:50.869818Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:51:50.869818Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Length Controlled Generation for Black-box LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Bing Qin, Kun Zhu, Lei Huang, Tat-Seng Chua, Weihong Zhong, Wenjie Wang, Xiaocheng Feng, Yuxuan Gu","submitted_at":"2024-12-19T09:07:38Z","abstract_excerpt":"Large language models (LLMs) have demonstrated impressive instruction following capabilities, while still struggling to accurately manage the length of the generated text, which is a fundamental requirement in many real-world applications. Existing length control methods involve fine-tuning the parameters of LLMs, which is inefficient and suboptimal for practical use. In this paper, we propose a novel iterative sampling framework for text length control, integrating the Metropolis-Hastings algorithm with an importance sampling acceleration strategy. This framework efficiently and reliably regu"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.14656","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/2412.14656/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":"2412.14656","created_at":"2026-07-05T09:51:50.869872+00:00"},{"alias_kind":"arxiv_version","alias_value":"2412.14656v1","created_at":"2026-07-05T09:51:50.869872+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.14656","created_at":"2026-07-05T09:51:50.869872+00:00"},{"alias_kind":"pith_short_12","alias_value":"NZD3JL7YRQA4","created_at":"2026-07-05T09:51:50.869872+00:00"},{"alias_kind":"pith_short_16","alias_value":"NZD3JL7YRQA4WSHG","created_at":"2026-07-05T09:51:50.869872+00:00"},{"alias_kind":"pith_short_8","alias_value":"NZD3JL7Y","created_at":"2026-07-05T09:51:50.869872+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2512.06938","citing_title":"Progress Ratio Embeddings: An Impatience Signal for Robust Length Control in Neural Text Generation","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2604.27039","citing_title":"Length Value Model: Scalable Value Pretraining for Token-Level Length Modeling","ref_index":7,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/NZD3JL7YRQA4WSHGGYAPV6SIKX","json":"https://pith.science/pith/NZD3JL7YRQA4WSHGGYAPV6SIKX.json","graph_json":"https://pith.science/api/pith-number/NZD3JL7YRQA4WSHGGYAPV6SIKX/graph.json","events_json":"https://pith.science/api/pith-number/NZD3JL7YRQA4WSHGGYAPV6SIKX/events.json","paper":"https://pith.science/paper/NZD3JL7Y"},"agent_actions":{"view_html":"https://pith.science/pith/NZD3JL7YRQA4WSHGGYAPV6SIKX","download_json":"https://pith.science/pith/NZD3JL7YRQA4WSHGGYAPV6SIKX.json","view_paper":"https://pith.science/paper/NZD3JL7Y","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2412.14656&json=true","fetch_graph":"https://pith.science/api/pith-number/NZD3JL7YRQA4WSHGGYAPV6SIKX/graph.json","fetch_events":"https://pith.science/api/pith-number/NZD3JL7YRQA4WSHGGYAPV6SIKX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NZD3JL7YRQA4WSHGGYAPV6SIKX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NZD3JL7YRQA4WSHGGYAPV6SIKX/action/storage_attestation","attest_author":"https://pith.science/pith/NZD3JL7YRQA4WSHGGYAPV6SIKX/action/author_attestation","sign_citation":"https://pith.science/pith/NZD3JL7YRQA4WSHGGYAPV6SIKX/action/citation_signature","submit_replication":"https://pith.science/pith/NZD3JL7YRQA4WSHGGYAPV6SIKX/action/replication_record"}},"created_at":"2026-07-05T09:51:50.869872+00:00","updated_at":"2026-07-05T09:51:50.869872+00:00"}