{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:6OXAXF5WGC227VSXW3QMQ3LYSR","short_pith_number":"pith:6OXAXF5W","schema_version":"1.0","canonical_sha256":"f3ae0b97b630b5afd657b6e0c86d78945804a4871a94adea803680b65b17cf2f","source":{"kind":"arxiv","id":"2502.01126","version":1},"attestation_state":"computed","paper":{"title":"Language Models Prefer What They Know: Relative Confidence Estimation via Confidence Preferences","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Ananya Kumar, Percy Liang, Vaishnavi Shrivastava","submitted_at":"2025-02-03T07:43:27Z","abstract_excerpt":"Language models (LMs) should provide reliable confidence estimates to help users detect mistakes in their outputs and defer to human experts when necessary. Asking a language model to assess its confidence (\"Score your confidence from 0-1.\") is a natural way of evaluating its uncertainty. However, models struggle to provide absolute assessments of confidence (i.e. judging confidence in answering a question independent of other questions) and the coarse-grained scores they produce are not useful for evaluating the correctness of their answers. We propose relative confidence estimation, where we"},"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":"2502.01126","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-02-03T07:43:27Z","cross_cats_sorted":[],"title_canon_sha256":"b36f6b550e8e1a8fa4b3e42c0747e3f0f9854854041e8886d3688f1286b1b3a3","abstract_canon_sha256":"44668db419f0887b09fa985763d4e19ec2f95d4dee87987956e61bf134d196d2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:08:43.162127Z","signature_b64":"OHwYBNonpWhIxsSE/AovQjN76oVi76Vq9WlP/CDk/M1cUHJ7WZOpYWgRloM7HmzpJMApEppm1VfSeeXKNw4OBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f3ae0b97b630b5afd657b6e0c86d78945804a4871a94adea803680b65b17cf2f","last_reissued_at":"2026-07-05T10:08:43.161692Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:08:43.161692Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Language Models Prefer What They Know: Relative Confidence Estimation via Confidence Preferences","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Ananya Kumar, Percy Liang, Vaishnavi Shrivastava","submitted_at":"2025-02-03T07:43:27Z","abstract_excerpt":"Language models (LMs) should provide reliable confidence estimates to help users detect mistakes in their outputs and defer to human experts when necessary. Asking a language model to assess its confidence (\"Score your confidence from 0-1.\") is a natural way of evaluating its uncertainty. However, models struggle to provide absolute assessments of confidence (i.e. judging confidence in answering a question independent of other questions) and the coarse-grained scores they produce are not useful for evaluating the correctness of their answers. We propose relative confidence estimation, where we"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.01126","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/2502.01126/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":"2502.01126","created_at":"2026-07-05T10:08:43.161748+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.01126v1","created_at":"2026-07-05T10:08:43.161748+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.01126","created_at":"2026-07-05T10:08:43.161748+00:00"},{"alias_kind":"pith_short_12","alias_value":"6OXAXF5WGC22","created_at":"2026-07-05T10:08:43.161748+00:00"},{"alias_kind":"pith_short_16","alias_value":"6OXAXF5WGC227VSX","created_at":"2026-07-05T10:08:43.161748+00:00"},{"alias_kind":"pith_short_8","alias_value":"6OXAXF5W","created_at":"2026-07-05T10:08:43.161748+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.17312","citing_title":"Quantifying Consistency in LLM Logical Reasoning via Structural Uncertainty","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2408.09049","citing_title":"Inertia in Moral and Value Judgments of Large Language Models","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2604.11801","citing_title":"CLSGen: A Dual-Head Fine-Tuning Framework for Joint Probabilistic Classification and Verbalized Explanation","ref_index":9,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6OXAXF5WGC227VSXW3QMQ3LYSR","json":"https://pith.science/pith/6OXAXF5WGC227VSXW3QMQ3LYSR.json","graph_json":"https://pith.science/api/pith-number/6OXAXF5WGC227VSXW3QMQ3LYSR/graph.json","events_json":"https://pith.science/api/pith-number/6OXAXF5WGC227VSXW3QMQ3LYSR/events.json","paper":"https://pith.science/paper/6OXAXF5W"},"agent_actions":{"view_html":"https://pith.science/pith/6OXAXF5WGC227VSXW3QMQ3LYSR","download_json":"https://pith.science/pith/6OXAXF5WGC227VSXW3QMQ3LYSR.json","view_paper":"https://pith.science/paper/6OXAXF5W","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.01126&json=true","fetch_graph":"https://pith.science/api/pith-number/6OXAXF5WGC227VSXW3QMQ3LYSR/graph.json","fetch_events":"https://pith.science/api/pith-number/6OXAXF5WGC227VSXW3QMQ3LYSR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6OXAXF5WGC227VSXW3QMQ3LYSR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6OXAXF5WGC227VSXW3QMQ3LYSR/action/storage_attestation","attest_author":"https://pith.science/pith/6OXAXF5WGC227VSXW3QMQ3LYSR/action/author_attestation","sign_citation":"https://pith.science/pith/6OXAXF5WGC227VSXW3QMQ3LYSR/action/citation_signature","submit_replication":"https://pith.science/pith/6OXAXF5WGC227VSXW3QMQ3LYSR/action/replication_record"}},"created_at":"2026-07-05T10:08:43.161748+00:00","updated_at":"2026-07-05T10:08:43.161748+00:00"}