{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:7I53SCW7FL43BGEXI5AC7ARG7Q","short_pith_number":"pith:7I53SCW7","schema_version":"1.0","canonical_sha256":"fa3bb90adf2af9b0989747402f8226fc2bbd2146ed76ef2086bf2108f97d154a","source":{"kind":"arxiv","id":"2607.01240","version":1},"attestation_state":"computed","paper":{"title":"Prompt Framing Distorts Count-Based Evaluation of LLM Error Detection: Evidence from Numeric Anchoring","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Dekun Yang","submitted_at":"2026-05-03T12:21:42Z","abstract_excerpt":"Count-based F1 is widely used as a proxy for LLM error-detection quality, but this paper shows that it can rise dramatically without a corresponding improvement in span localization, a gap termed F1 Inflation. The paper introduces ErrorBench, a controlled stress-test protocol for prompt-induced count distortion. ErrorBench evaluates six contemporary LLMs under five prompt conditions over 4,290 responses from 143 CoNLL-2014 passages. Under CoNLL-2014 M2-style scoring, anchored prompts produce up to 0.79 points of F1 Inflation, and up to 0.96 under strict matching. A 100-passage replication usin"},"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":"2607.01240","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-05-03T12:21:42Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"069a01dbaaeb97985061c7e656eca969ec5ef18fcb4e4528e2d62f374ab2c8d6","abstract_canon_sha256":"0ca7c8aa5a9ed30eea1dcd7cda2c1b390ddf92344f25f933c08e9c518c2d6839"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-03T00:16:55.070472Z","signature_b64":"AiNRfyph/OAeVTDXf/3RwWajnaEFVvWsGrmJwxnHVUE8vAwRtBPSSbAmwrysKxTDnvEWI5LV47TbD4TF8r3sAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fa3bb90adf2af9b0989747402f8226fc2bbd2146ed76ef2086bf2108f97d154a","last_reissued_at":"2026-07-03T00:16:55.070116Z","signature_status":"signed_v1","first_computed_at":"2026-07-03T00:16:55.070116Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Prompt Framing Distorts Count-Based Evaluation of LLM Error Detection: Evidence from Numeric Anchoring","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Dekun Yang","submitted_at":"2026-05-03T12:21:42Z","abstract_excerpt":"Count-based F1 is widely used as a proxy for LLM error-detection quality, but this paper shows that it can rise dramatically without a corresponding improvement in span localization, a gap termed F1 Inflation. The paper introduces ErrorBench, a controlled stress-test protocol for prompt-induced count distortion. ErrorBench evaluates six contemporary LLMs under five prompt conditions over 4,290 responses from 143 CoNLL-2014 passages. Under CoNLL-2014 M2-style scoring, anchored prompts produce up to 0.79 points of F1 Inflation, and up to 0.96 under strict matching. A 100-passage replication usin"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.01240","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/2607.01240/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":"2607.01240","created_at":"2026-07-03T00:16:55.070176+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.01240v1","created_at":"2026-07-03T00:16:55.070176+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.01240","created_at":"2026-07-03T00:16:55.070176+00:00"},{"alias_kind":"pith_short_12","alias_value":"7I53SCW7FL43","created_at":"2026-07-03T00:16:55.070176+00:00"},{"alias_kind":"pith_short_16","alias_value":"7I53SCW7FL43BGEX","created_at":"2026-07-03T00:16:55.070176+00:00"},{"alias_kind":"pith_short_8","alias_value":"7I53SCW7","created_at":"2026-07-03T00:16:55.070176+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/7I53SCW7FL43BGEXI5AC7ARG7Q","json":"https://pith.science/pith/7I53SCW7FL43BGEXI5AC7ARG7Q.json","graph_json":"https://pith.science/api/pith-number/7I53SCW7FL43BGEXI5AC7ARG7Q/graph.json","events_json":"https://pith.science/api/pith-number/7I53SCW7FL43BGEXI5AC7ARG7Q/events.json","paper":"https://pith.science/paper/7I53SCW7"},"agent_actions":{"view_html":"https://pith.science/pith/7I53SCW7FL43BGEXI5AC7ARG7Q","download_json":"https://pith.science/pith/7I53SCW7FL43BGEXI5AC7ARG7Q.json","view_paper":"https://pith.science/paper/7I53SCW7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.01240&json=true","fetch_graph":"https://pith.science/api/pith-number/7I53SCW7FL43BGEXI5AC7ARG7Q/graph.json","fetch_events":"https://pith.science/api/pith-number/7I53SCW7FL43BGEXI5AC7ARG7Q/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7I53SCW7FL43BGEXI5AC7ARG7Q/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7I53SCW7FL43BGEXI5AC7ARG7Q/action/storage_attestation","attest_author":"https://pith.science/pith/7I53SCW7FL43BGEXI5AC7ARG7Q/action/author_attestation","sign_citation":"https://pith.science/pith/7I53SCW7FL43BGEXI5AC7ARG7Q/action/citation_signature","submit_replication":"https://pith.science/pith/7I53SCW7FL43BGEXI5AC7ARG7Q/action/replication_record"}},"created_at":"2026-07-03T00:16:55.070176+00:00","updated_at":"2026-07-03T00:16:55.070176+00:00"}