{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:BDEUCIDKOGOTYUAD7WUPRXFHOH","short_pith_number":"pith:BDEUCIDK","schema_version":"1.0","canonical_sha256":"08c941206a719d3c5003fda8f8dca771d963fc6c3142a4e0419349a53afd505c","source":{"kind":"arxiv","id":"2507.10419","version":3},"attestation_state":"computed","paper":{"title":"Multiple Choice Learning of Low-Rank Adapters for Language Modeling","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Andrei Bursuc, Ga\\\"el Richard, Hugo Malard, Mathieu Fontaine, Patrick P\\'erez, Slim Essid, Victor Letzelter","submitted_at":"2025-07-14T16:00:51Z","abstract_excerpt":"We propose LoRA-MCL, a training scheme that extends next-token prediction in language models with a method designed to decode diverse, plausible sentence continuations at inference time. Traditional language modeling is an intrinsically ill-posed problem: given a context, multiple futures may be equally plausible. Our approach leverages Multiple Choice Learning (MCL) and the winner-takes-all loss to efficiently handle ambiguity through Low-Rank Adaptation. We provide a theoretical interpretation of applying MCL to language modeling, assuming the data is generated from a mixture of distribution"},"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":"2507.10419","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.LG","submitted_at":"2025-07-14T16:00:51Z","cross_cats_sorted":["cs.AI","cs.CL","stat.ML"],"title_canon_sha256":"f270359a9926b8cadbf02133cb4cd830c159433c03c078e26fdd75e5f60ca57c","abstract_canon_sha256":"70c46c99e6793bea39d11b113b9d4e8cd5497fa3d60421181c55913a14359caa"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-03T01:05:44.351797Z","signature_b64":"bU5QFpmCB4RsvlMpS2F8AJJNJ62Y9OgMuSkCOAiPzM4N4Pr+aiDoOP00EgWR8p1GkWKo2W2gJI/vxcwTbCnpCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"08c941206a719d3c5003fda8f8dca771d963fc6c3142a4e0419349a53afd505c","last_reissued_at":"2026-06-03T01:05:44.351301Z","signature_status":"signed_v1","first_computed_at":"2026-06-03T01:05:44.351301Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Multiple Choice Learning of Low-Rank Adapters for Language Modeling","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Andrei Bursuc, Ga\\\"el Richard, Hugo Malard, Mathieu Fontaine, Patrick P\\'erez, Slim Essid, Victor Letzelter","submitted_at":"2025-07-14T16:00:51Z","abstract_excerpt":"We propose LoRA-MCL, a training scheme that extends next-token prediction in language models with a method designed to decode diverse, plausible sentence continuations at inference time. Traditional language modeling is an intrinsically ill-posed problem: given a context, multiple futures may be equally plausible. Our approach leverages Multiple Choice Learning (MCL) and the winner-takes-all loss to efficiently handle ambiguity through Low-Rank Adaptation. We provide a theoretical interpretation of applying MCL to language modeling, assuming the data is generated from a mixture of distribution"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.10419","kind":"arxiv","version":3},"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/2507.10419/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":"2507.10419","created_at":"2026-06-03T01:05:44.351373+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.10419v3","created_at":"2026-06-03T01:05:44.351373+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.10419","created_at":"2026-06-03T01:05:44.351373+00:00"},{"alias_kind":"pith_short_12","alias_value":"BDEUCIDKOGOT","created_at":"2026-06-03T01:05:44.351373+00:00"},{"alias_kind":"pith_short_16","alias_value":"BDEUCIDKOGOTYUAD","created_at":"2026-06-03T01:05:44.351373+00:00"},{"alias_kind":"pith_short_8","alias_value":"BDEUCIDK","created_at":"2026-06-03T01:05:44.351373+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/BDEUCIDKOGOTYUAD7WUPRXFHOH","json":"https://pith.science/pith/BDEUCIDKOGOTYUAD7WUPRXFHOH.json","graph_json":"https://pith.science/api/pith-number/BDEUCIDKOGOTYUAD7WUPRXFHOH/graph.json","events_json":"https://pith.science/api/pith-number/BDEUCIDKOGOTYUAD7WUPRXFHOH/events.json","paper":"https://pith.science/paper/BDEUCIDK"},"agent_actions":{"view_html":"https://pith.science/pith/BDEUCIDKOGOTYUAD7WUPRXFHOH","download_json":"https://pith.science/pith/BDEUCIDKOGOTYUAD7WUPRXFHOH.json","view_paper":"https://pith.science/paper/BDEUCIDK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.10419&json=true","fetch_graph":"https://pith.science/api/pith-number/BDEUCIDKOGOTYUAD7WUPRXFHOH/graph.json","fetch_events":"https://pith.science/api/pith-number/BDEUCIDKOGOTYUAD7WUPRXFHOH/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BDEUCIDKOGOTYUAD7WUPRXFHOH/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BDEUCIDKOGOTYUAD7WUPRXFHOH/action/storage_attestation","attest_author":"https://pith.science/pith/BDEUCIDKOGOTYUAD7WUPRXFHOH/action/author_attestation","sign_citation":"https://pith.science/pith/BDEUCIDKOGOTYUAD7WUPRXFHOH/action/citation_signature","submit_replication":"https://pith.science/pith/BDEUCIDKOGOTYUAD7WUPRXFHOH/action/replication_record"}},"created_at":"2026-06-03T01:05:44.351373+00:00","updated_at":"2026-06-03T01:05:44.351373+00:00"}