{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:APZ4THPG3LCRAWE3WO25XN3ETR","short_pith_number":"pith:APZ4THPG","schema_version":"1.0","canonical_sha256":"03f3c99de6dac510589bb3b5dbb7649c59b2415ed940550d85cc4c348df201af","source":{"kind":"arxiv","id":"2404.10960","version":1},"attestation_state":"computed","paper":{"title":"Uncertainty-Based Abstention in LLMs Improves Safety and Reduces Hallucinations","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Christian Tomani, Daniel Cremers, Ivan Evtimov, Kamalika Chaudhuri, Mark Ibrahim","submitted_at":"2024-04-16T23:56:38Z","abstract_excerpt":"A major barrier towards the practical deployment of large language models (LLMs) is their lack of reliability. Three situations where this is particularly apparent are correctness, hallucinations when given unanswerable questions, and safety. In all three cases, models should ideally abstain from responding, much like humans, whose ability to understand uncertainty makes us refrain from answering questions we don't know. Inspired by analogous approaches in classification, this study explores the feasibility and efficacy of abstaining while uncertain in the context of LLMs within the domain of "},"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":"2404.10960","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-04-16T23:56:38Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"9ce1b29991f83eab62553a0d53b3f316f7395ec137d6ae67b1ae48d5305a5ee0","abstract_canon_sha256":"d26f1c24e82098565c6a311e93830f04066c91550e832453477e8de3bf650541"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:09:01.384258Z","signature_b64":"JNVdS1TRhamLUQSIncf4aFL89vl8tUNFunxQfqXr5wiLLZToM0HhDG6MKeLB90I/tqxZ9Pq1KxIrKuz9URTYDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"03f3c99de6dac510589bb3b5dbb7649c59b2415ed940550d85cc4c348df201af","last_reissued_at":"2026-07-05T08:09:01.383762Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:09:01.383762Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Uncertainty-Based Abstention in LLMs Improves Safety and Reduces Hallucinations","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Christian Tomani, Daniel Cremers, Ivan Evtimov, Kamalika Chaudhuri, Mark Ibrahim","submitted_at":"2024-04-16T23:56:38Z","abstract_excerpt":"A major barrier towards the practical deployment of large language models (LLMs) is their lack of reliability. Three situations where this is particularly apparent are correctness, hallucinations when given unanswerable questions, and safety. In all three cases, models should ideally abstain from responding, much like humans, whose ability to understand uncertainty makes us refrain from answering questions we don't know. Inspired by analogous approaches in classification, this study explores the feasibility and efficacy of abstaining while uncertain in the context of LLMs within the domain of "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2404.10960","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/2404.10960/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":"2404.10960","created_at":"2026-07-05T08:09:01.383821+00:00"},{"alias_kind":"arxiv_version","alias_value":"2404.10960v1","created_at":"2026-07-05T08:09:01.383821+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2404.10960","created_at":"2026-07-05T08:09:01.383821+00:00"},{"alias_kind":"pith_short_12","alias_value":"APZ4THPG3LCR","created_at":"2026-07-05T08:09:01.383821+00:00"},{"alias_kind":"pith_short_16","alias_value":"APZ4THPG3LCRAWE3","created_at":"2026-07-05T08:09:01.383821+00:00"},{"alias_kind":"pith_short_8","alias_value":"APZ4THPG","created_at":"2026-07-05T08:09:01.383821+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":9,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.22728","citing_title":"When Confidence Takes the Wrong Path: Diagnosing Retrieval-State Lock-In in RAG","ref_index":57,"is_internal_anchor":false},{"citing_arxiv_id":"2606.04661","citing_title":"CRAFT: Cost-aware Refinement And Front-aware Tuning of Prompts","ref_index":196,"is_internal_anchor":false},{"citing_arxiv_id":"2606.27922","citing_title":"Reflect-R1: Evidence-Driven Reflection for Self-Correction in Long Video Understanding","ref_index":30,"is_internal_anchor":false},{"citing_arxiv_id":"2606.27922","citing_title":"Reflect-R1: Evidence-Driven Reflection for Self-Correction in Long Video Understanding","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2606.27922","citing_title":"Reflect-R1: Evidence-Driven Reflection for Self-Correction in Long Video Understanding","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2603.22161","citing_title":"Causal Evidence that Language Models use Confidence to Drive Behavior","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09278","citing_title":"EquiMem: Calibrating Shared Memory in Multi-Agent Debate via Game-Theoretic Equilibrium","ref_index":65,"is_internal_anchor":false},{"citing_arxiv_id":"2604.06714","citing_title":"Steering the Verifiability of Multimodal AI Hallucinations","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2604.16686","citing_title":"No-Worse Context-Aware Decoding: Preventing Neutral Regression in Context-Conditioned Generation","ref_index":22,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/APZ4THPG3LCRAWE3WO25XN3ETR","json":"https://pith.science/pith/APZ4THPG3LCRAWE3WO25XN3ETR.json","graph_json":"https://pith.science/api/pith-number/APZ4THPG3LCRAWE3WO25XN3ETR/graph.json","events_json":"https://pith.science/api/pith-number/APZ4THPG3LCRAWE3WO25XN3ETR/events.json","paper":"https://pith.science/paper/APZ4THPG"},"agent_actions":{"view_html":"https://pith.science/pith/APZ4THPG3LCRAWE3WO25XN3ETR","download_json":"https://pith.science/pith/APZ4THPG3LCRAWE3WO25XN3ETR.json","view_paper":"https://pith.science/paper/APZ4THPG","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2404.10960&json=true","fetch_graph":"https://pith.science/api/pith-number/APZ4THPG3LCRAWE3WO25XN3ETR/graph.json","fetch_events":"https://pith.science/api/pith-number/APZ4THPG3LCRAWE3WO25XN3ETR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/APZ4THPG3LCRAWE3WO25XN3ETR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/APZ4THPG3LCRAWE3WO25XN3ETR/action/storage_attestation","attest_author":"https://pith.science/pith/APZ4THPG3LCRAWE3WO25XN3ETR/action/author_attestation","sign_citation":"https://pith.science/pith/APZ4THPG3LCRAWE3WO25XN3ETR/action/citation_signature","submit_replication":"https://pith.science/pith/APZ4THPG3LCRAWE3WO25XN3ETR/action/replication_record"}},"created_at":"2026-07-05T08:09:01.383821+00:00","updated_at":"2026-07-05T08:09:01.383821+00:00"}