{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:6D4PDLVD6K3SZ2GUV3STYGJT2F","short_pith_number":"pith:6D4PDLVD","schema_version":"1.0","canonical_sha256":"f0f8f1aea3f2b72ce8d4aee53c1933d16e9cf396b93b38048f07b8f812bccb55","source":{"kind":"arxiv","id":"2404.10774","version":2},"attestation_state":"computed","paper":{"title":"MiniCheck: Efficient Fact-Checking of LLMs on Grounding Documents","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Greg Durrett, Liyan Tang, Philippe Laban","submitted_at":"2024-04-16T17:59:10Z","abstract_excerpt":"Recognizing if LLM output can be grounded in evidence is central to many tasks in NLP: retrieval-augmented generation, summarization, document-grounded dialogue, and more. Current approaches to this kind of fact-checking are based on verifying each piece of a model generation against potential evidence using an LLM. However, this process can be very computationally expensive, requiring many calls to a model to check a single response. In this work, we show how to build small fact-checking models that have GPT-4-level performance but for 400x lower cost. We do this by constructing synthetic tra"},"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":"2404.10774","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-04-16T17:59:10Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"082a6c6e37a073a6f12461e7449c94f827ba2c9876fdf2d767ee6714f5e93120","abstract_canon_sha256":"2ea4247f16621dc94410d8b339abfaace5d29d34baab5637d0c67beef0e59403"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:13:51.964895Z","signature_b64":"k7seG6dUeVseCeVLd9GhiVCznhuWupg9Pz+tbROjcOVif7WWJBeruPOst12wC8efI+56xaFDwDABrbg6rYzlDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f0f8f1aea3f2b72ce8d4aee53c1933d16e9cf396b93b38048f07b8f812bccb55","last_reissued_at":"2026-07-05T09:13:51.964417Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:13:51.964417Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MiniCheck: Efficient Fact-Checking of LLMs on Grounding Documents","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Greg Durrett, Liyan Tang, Philippe Laban","submitted_at":"2024-04-16T17:59:10Z","abstract_excerpt":"Recognizing if LLM output can be grounded in evidence is central to many tasks in NLP: retrieval-augmented generation, summarization, document-grounded dialogue, and more. Current approaches to this kind of fact-checking are based on verifying each piece of a model generation against potential evidence using an LLM. However, this process can be very computationally expensive, requiring many calls to a model to check a single response. In this work, we show how to build small fact-checking models that have GPT-4-level performance but for 400x lower cost. We do this by constructing synthetic tra"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2404.10774","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/2404.10774/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":"2404.10774","created_at":"2026-07-05T09:13:51.964477+00:00"},{"alias_kind":"arxiv_version","alias_value":"2404.10774v2","created_at":"2026-07-05T09:13:51.964477+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2404.10774","created_at":"2026-07-05T09:13:51.964477+00:00"},{"alias_kind":"pith_short_12","alias_value":"6D4PDLVD6K3S","created_at":"2026-07-05T09:13:51.964477+00:00"},{"alias_kind":"pith_short_16","alias_value":"6D4PDLVD6K3SZ2GU","created_at":"2026-07-05T09:13:51.964477+00:00"},{"alias_kind":"pith_short_8","alias_value":"6D4PDLVD","created_at":"2026-07-05T09:13:51.964477+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":12,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.08268","citing_title":"Different Teachers, Different Capabilities: Sub-1B On-Device Distillation for Structured Text Enrichment","ref_index":41,"is_internal_anchor":true},{"citing_arxiv_id":"2606.13216","citing_title":"Layer-Resolved Optimal Transport for Hallucination Detection in NMT and Abstractive Summarization","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2606.12807","citing_title":"Detect, Remask, Repair: Diffusion Editing for Faithful Summarization of Evolving Contexts","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2606.08158","citing_title":"Constrained Paraphrase Consistency for LLM Hallucination Detection","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2606.08157","citing_title":"Cross Paraphrastic Invariance Learning for Hallucination Detection","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2606.29713","citing_title":"SEVA: Self-Evolving Verification Agent with Process Reward for Fact Attribution","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2605.02443","citing_title":"HalluScan: A Systematic Benchmark for Detecting and Mitigating Hallucinations in Instruction-Following LLMs","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2504.02181","citing_title":"A Survey of Scaling in Large Language Model Reasoning","ref_index":195,"is_internal_anchor":false},{"citing_arxiv_id":"2604.11036","citing_title":"Uncertainty-Aware Web-Conditioned Scientific Fact-Checking","ref_index":16,"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":19,"is_internal_anchor":false},{"citing_arxiv_id":"2604.15945","citing_title":"RAGognizer: Hallucination-Aware Fine-Tuning via Detection Head Integration","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2604.20131","citing_title":"Whose Story Gets Told? Positionality and Bias in LLM Summaries of Life Narratives","ref_index":162,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6D4PDLVD6K3SZ2GUV3STYGJT2F","json":"https://pith.science/pith/6D4PDLVD6K3SZ2GUV3STYGJT2F.json","graph_json":"https://pith.science/api/pith-number/6D4PDLVD6K3SZ2GUV3STYGJT2F/graph.json","events_json":"https://pith.science/api/pith-number/6D4PDLVD6K3SZ2GUV3STYGJT2F/events.json","paper":"https://pith.science/paper/6D4PDLVD"},"agent_actions":{"view_html":"https://pith.science/pith/6D4PDLVD6K3SZ2GUV3STYGJT2F","download_json":"https://pith.science/pith/6D4PDLVD6K3SZ2GUV3STYGJT2F.json","view_paper":"https://pith.science/paper/6D4PDLVD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2404.10774&json=true","fetch_graph":"https://pith.science/api/pith-number/6D4PDLVD6K3SZ2GUV3STYGJT2F/graph.json","fetch_events":"https://pith.science/api/pith-number/6D4PDLVD6K3SZ2GUV3STYGJT2F/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6D4PDLVD6K3SZ2GUV3STYGJT2F/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6D4PDLVD6K3SZ2GUV3STYGJT2F/action/storage_attestation","attest_author":"https://pith.science/pith/6D4PDLVD6K3SZ2GUV3STYGJT2F/action/author_attestation","sign_citation":"https://pith.science/pith/6D4PDLVD6K3SZ2GUV3STYGJT2F/action/citation_signature","submit_replication":"https://pith.science/pith/6D4PDLVD6K3SZ2GUV3STYGJT2F/action/replication_record"}},"created_at":"2026-07-05T09:13:51.964477+00:00","updated_at":"2026-07-05T09:13:51.964477+00:00"}