{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:PELZ7BNSKPTBSDCSFVD32RTD2D","short_pith_number":"pith:PELZ7BNS","schema_version":"1.0","canonical_sha256":"79179f85b253e6190c522d47bd4663d0d2748f31d64f2f6fa415b704bff3fbda","source":{"kind":"arxiv","id":"2605.29797","version":1},"attestation_state":"computed","paper":{"title":"Metric-Dependent Annotation Saturation for Learning from Label Distributions","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Guneet Kohli","submitted_at":"2026-05-28T11:46:51Z","abstract_excerpt":"When annotators disagree on a label, the disagreement itself carries signal -- and the number of annotators needed to capture it depends on the evaluation metric. We fine-tune NLI models on label distributions subsampled from ChaosNLI, a dataset providing 100 independent annotator judgments per item, and identify metric-dependent saturation. In our 3-class NLI setting, entropy correlation -- whether the model identifies which items elicit disagreement -- requires N ~ 20-50 annotators to converge, while distributional match (KL divergence) saturates by N ~ 10 (87-95% of improvement across five "},"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":"2605.29797","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-05-28T11:46:51Z","cross_cats_sorted":[],"title_canon_sha256":"0f85e9c87a788a174cadc7aacce64a546b6f78f2d9e0503d19ab1eaa5611df32","abstract_canon_sha256":"3681076af8ac9aac758bc024ee3328627ec0321acd83cb8209f16170514f8793"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-29T02:05:52.871839Z","signature_b64":"MazzEYOpVkTwbkHPStB3+ZYbEIVOkQA6BKcAs/JLqWw3mNKAWh4t15Lx47KVFs6gtIqD+42/c1FUym/1TOFIDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"79179f85b253e6190c522d47bd4663d0d2748f31d64f2f6fa415b704bff3fbda","last_reissued_at":"2026-05-29T02:05:52.870927Z","signature_status":"signed_v1","first_computed_at":"2026-05-29T02:05:52.870927Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Metric-Dependent Annotation Saturation for Learning from Label Distributions","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Guneet Kohli","submitted_at":"2026-05-28T11:46:51Z","abstract_excerpt":"When annotators disagree on a label, the disagreement itself carries signal -- and the number of annotators needed to capture it depends on the evaluation metric. We fine-tune NLI models on label distributions subsampled from ChaosNLI, a dataset providing 100 independent annotator judgments per item, and identify metric-dependent saturation. In our 3-class NLI setting, entropy correlation -- whether the model identifies which items elicit disagreement -- requires N ~ 20-50 annotators to converge, while distributional match (KL divergence) saturates by N ~ 10 (87-95% of improvement across five "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.29797","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/2605.29797/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":"2605.29797","created_at":"2026-05-29T02:05:52.871079+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.29797v1","created_at":"2026-05-29T02:05:52.871079+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.29797","created_at":"2026-05-29T02:05:52.871079+00:00"},{"alias_kind":"pith_short_12","alias_value":"PELZ7BNSKPTB","created_at":"2026-05-29T02:05:52.871079+00:00"},{"alias_kind":"pith_short_16","alias_value":"PELZ7BNSKPTBSDCS","created_at":"2026-05-29T02:05:52.871079+00:00"},{"alias_kind":"pith_short_8","alias_value":"PELZ7BNS","created_at":"2026-05-29T02:05:52.871079+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/PELZ7BNSKPTBSDCSFVD32RTD2D","json":"https://pith.science/pith/PELZ7BNSKPTBSDCSFVD32RTD2D.json","graph_json":"https://pith.science/api/pith-number/PELZ7BNSKPTBSDCSFVD32RTD2D/graph.json","events_json":"https://pith.science/api/pith-number/PELZ7BNSKPTBSDCSFVD32RTD2D/events.json","paper":"https://pith.science/paper/PELZ7BNS"},"agent_actions":{"view_html":"https://pith.science/pith/PELZ7BNSKPTBSDCSFVD32RTD2D","download_json":"https://pith.science/pith/PELZ7BNSKPTBSDCSFVD32RTD2D.json","view_paper":"https://pith.science/paper/PELZ7BNS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.29797&json=true","fetch_graph":"https://pith.science/api/pith-number/PELZ7BNSKPTBSDCSFVD32RTD2D/graph.json","fetch_events":"https://pith.science/api/pith-number/PELZ7BNSKPTBSDCSFVD32RTD2D/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PELZ7BNSKPTBSDCSFVD32RTD2D/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PELZ7BNSKPTBSDCSFVD32RTD2D/action/storage_attestation","attest_author":"https://pith.science/pith/PELZ7BNSKPTBSDCSFVD32RTD2D/action/author_attestation","sign_citation":"https://pith.science/pith/PELZ7BNSKPTBSDCSFVD32RTD2D/action/citation_signature","submit_replication":"https://pith.science/pith/PELZ7BNSKPTBSDCSFVD32RTD2D/action/replication_record"}},"created_at":"2026-05-29T02:05:52.871079+00:00","updated_at":"2026-05-29T02:05:52.871079+00:00"}