{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:4W5AGCPQNWWRIG62KUDY3HCOI7","short_pith_number":"pith:4W5AGCPQ","schema_version":"1.0","canonical_sha256":"e5ba0309f06dad141bda55078d9c4e47c49c7abe754921ae8817e265e510e504","source":{"kind":"arxiv","id":"2506.09659","version":1},"attestation_state":"computed","paper":{"title":"Intent Factored Generation: Unleashing the Diversity in Your Language Model","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.AI","authors_text":"Danijela Horak, Eltayeb Ahmed, Jakob N. Foerster, Martha Elliott, Uljad Berdica","submitted_at":"2025-06-11T12:26:45Z","abstract_excerpt":"Obtaining multiple meaningfully diverse, high quality samples from Large Language Models for a fixed prompt remains an open challenge. Current methods for increasing diversity often only operate at the token-level, paraphrasing the same response. This is problematic because it leads to poor exploration on reasoning problems and to unengaging, repetitive conversational agents. To address this we propose Intent Factored Generation (IFG), factorising the sampling process into two stages. First, we sample a semantically dense intent, e.g., a summary or keywords. Second, we sample the final respons"},"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":"2506.09659","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.AI","submitted_at":"2025-06-11T12:26:45Z","cross_cats_sorted":["cs.CL","cs.LG"],"title_canon_sha256":"b6871d252b1727db642df0108d4a30ff6e81fea2e4c864263bc26e249f1cd884","abstract_canon_sha256":"95655d4c2145023b5ad346bd4c75b96390e8b229ab94742e9af56c8532fa5281"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:19:54.705979Z","signature_b64":"hC2x1Aa/kRTXQW06eausmgmKGFA8dExccT3hovfJdxLRzMBi7ne//w0vQ7bzVZiyhgyQNan76yGvJZDaTWCCDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e5ba0309f06dad141bda55078d9c4e47c49c7abe754921ae8817e265e510e504","last_reissued_at":"2026-07-05T11:19:54.705419Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:19:54.705419Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Intent Factored Generation: Unleashing the Diversity in Your Language Model","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.AI","authors_text":"Danijela Horak, Eltayeb Ahmed, Jakob N. Foerster, Martha Elliott, Uljad Berdica","submitted_at":"2025-06-11T12:26:45Z","abstract_excerpt":"Obtaining multiple meaningfully diverse, high quality samples from Large Language Models for a fixed prompt remains an open challenge. Current methods for increasing diversity often only operate at the token-level, paraphrasing the same response. This is problematic because it leads to poor exploration on reasoning problems and to unengaging, repetitive conversational agents. To address this we propose Intent Factored Generation (IFG), factorising the sampling process into two stages. First, we sample a semantically dense intent, e.g., a summary or keywords. Second, we sample the final respons"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.09659","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/2506.09659/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":"2506.09659","created_at":"2026-07-05T11:19:54.705482+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.09659v1","created_at":"2026-07-05T11:19:54.705482+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.09659","created_at":"2026-07-05T11:19:54.705482+00:00"},{"alias_kind":"pith_short_12","alias_value":"4W5AGCPQNWWR","created_at":"2026-07-05T11:19:54.705482+00:00"},{"alias_kind":"pith_short_16","alias_value":"4W5AGCPQNWWRIG62","created_at":"2026-07-05T11:19:54.705482+00:00"},{"alias_kind":"pith_short_8","alias_value":"4W5AGCPQ","created_at":"2026-07-05T11:19:54.705482+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.09995","citing_title":"Annotations Mitigate Post-Training Mode Collapse","ref_index":50,"is_internal_anchor":false},{"citing_arxiv_id":"2604.05859","citing_title":"When Do We Need LLMs? A Diagnostic for Language-Driven Bandits","ref_index":3,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/4W5AGCPQNWWRIG62KUDY3HCOI7","json":"https://pith.science/pith/4W5AGCPQNWWRIG62KUDY3HCOI7.json","graph_json":"https://pith.science/api/pith-number/4W5AGCPQNWWRIG62KUDY3HCOI7/graph.json","events_json":"https://pith.science/api/pith-number/4W5AGCPQNWWRIG62KUDY3HCOI7/events.json","paper":"https://pith.science/paper/4W5AGCPQ"},"agent_actions":{"view_html":"https://pith.science/pith/4W5AGCPQNWWRIG62KUDY3HCOI7","download_json":"https://pith.science/pith/4W5AGCPQNWWRIG62KUDY3HCOI7.json","view_paper":"https://pith.science/paper/4W5AGCPQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.09659&json=true","fetch_graph":"https://pith.science/api/pith-number/4W5AGCPQNWWRIG62KUDY3HCOI7/graph.json","fetch_events":"https://pith.science/api/pith-number/4W5AGCPQNWWRIG62KUDY3HCOI7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/4W5AGCPQNWWRIG62KUDY3HCOI7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/4W5AGCPQNWWRIG62KUDY3HCOI7/action/storage_attestation","attest_author":"https://pith.science/pith/4W5AGCPQNWWRIG62KUDY3HCOI7/action/author_attestation","sign_citation":"https://pith.science/pith/4W5AGCPQNWWRIG62KUDY3HCOI7/action/citation_signature","submit_replication":"https://pith.science/pith/4W5AGCPQNWWRIG62KUDY3HCOI7/action/replication_record"}},"created_at":"2026-07-05T11:19:54.705482+00:00","updated_at":"2026-07-05T11:19:54.705482+00:00"}