{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:LR3WB3LUT4MGYY3RXX2ZVGEBU2","short_pith_number":"pith:LR3WB3LU","schema_version":"1.0","canonical_sha256":"5c7760ed749f186c6371bdf59a9881a6a19ef0cfc30f330c97f32d509d028e62","source":{"kind":"arxiv","id":"2511.14117","version":2},"attestation_state":"computed","paper":{"title":"Distributions In, Distributions Out: The Case for Soft-Label Training","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Agamdeep Singh, Ashish Tiwari, Hosein Hasanbeig, Priyanshu Gupta","submitted_at":"2025-11-18T04:02:29Z","abstract_excerpt":"Supervised classifiers output a distribution over classes but are typically trained against a single label obtained by collapsing multiple annotators into a majority vote. On tasks where annotator disagreement reflects genuine ambiguity -- natural language inference, politeness, visually ambiguous categorization -- this collapse discards information and forces models to express uniform confidence on inputs where humans systematically disagree. We compare soft-label training, which uses the full annotation distribution as the target, against hard-label training across three datasets spanning vi"},"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":"2511.14117","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-11-18T04:02:29Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"ec25906516b61a27c9fe1709eed310d1284838600f2f48611dfc36b727a5bfcc","abstract_canon_sha256":"50749183e719f4d4d93fd468906ef1b421eeee004ebb3da0745f5b48cc445b09"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5c7760ed749f186c6371bdf59a9881a6a19ef0cfc30f330c97f32d509d028e62","last_reissued_at":"2026-07-31T01:08:53.793392Z","signature_status":"unsigned_v0","first_computed_at":"2026-07-31T01:08:53.793392Z"},"graph_snapshot":{"paper":{"title":"Distributions In, Distributions Out: The Case for Soft-Label Training","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Agamdeep Singh, Ashish Tiwari, Hosein Hasanbeig, Priyanshu Gupta","submitted_at":"2025-11-18T04:02:29Z","abstract_excerpt":"Supervised classifiers output a distribution over classes but are typically trained against a single label obtained by collapsing multiple annotators into a majority vote. On tasks where annotator disagreement reflects genuine ambiguity -- natural language inference, politeness, visually ambiguous categorization -- this collapse discards information and forces models to express uniform confidence on inputs where humans systematically disagree. We compare soft-label training, which uses the full annotation distribution as the target, against hard-label training across three datasets spanning vi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2511.14117","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/2511.14117/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":"2511.14117","created_at":"2026-07-31T01:08:53.796533+00:00"},{"alias_kind":"arxiv_version","alias_value":"2511.14117v2","created_at":"2026-07-31T01:08:53.796533+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2511.14117","created_at":"2026-07-31T01:08:53.796533+00:00"},{"alias_kind":"pith_short_12","alias_value":"LR3WB3LUT4MG","created_at":"2026-07-31T01:08:53.796533+00:00"},{"alias_kind":"pith_short_16","alias_value":"LR3WB3LUT4MGYY3R","created_at":"2026-07-31T01:08:53.796533+00:00"},{"alias_kind":"pith_short_8","alias_value":"LR3WB3LU","created_at":"2026-07-31T01:08:53.796533+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2605.29797","citing_title":"Metric-Dependent Annotation Saturation for Learning from Label Distributions","ref_index":3,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/LR3WB3LUT4MGYY3RXX2ZVGEBU2","json":"https://pith.science/pith/LR3WB3LUT4MGYY3RXX2ZVGEBU2.json","graph_json":"https://pith.science/api/pith-number/LR3WB3LUT4MGYY3RXX2ZVGEBU2/graph.json","events_json":"https://pith.science/api/pith-number/LR3WB3LUT4MGYY3RXX2ZVGEBU2/events.json","paper":"https://pith.science/paper/LR3WB3LU"},"agent_actions":{"view_html":"https://pith.science/pith/LR3WB3LUT4MGYY3RXX2ZVGEBU2","download_json":"https://pith.science/pith/LR3WB3LUT4MGYY3RXX2ZVGEBU2.json","view_paper":"https://pith.science/paper/LR3WB3LU","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2511.14117&json=true","fetch_graph":"https://pith.science/api/pith-number/LR3WB3LUT4MGYY3RXX2ZVGEBU2/graph.json","fetch_events":"https://pith.science/api/pith-number/LR3WB3LUT4MGYY3RXX2ZVGEBU2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LR3WB3LUT4MGYY3RXX2ZVGEBU2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LR3WB3LUT4MGYY3RXX2ZVGEBU2/action/storage_attestation","attest_author":"https://pith.science/pith/LR3WB3LUT4MGYY3RXX2ZVGEBU2/action/author_attestation","sign_citation":"https://pith.science/pith/LR3WB3LUT4MGYY3RXX2ZVGEBU2/action/citation_signature","submit_replication":"https://pith.science/pith/LR3WB3LUT4MGYY3RXX2ZVGEBU2/action/replication_record"}},"created_at":"2026-07-31T01:08:53.796533+00:00","updated_at":"2026-07-31T01:08:53.796533+00:00"}