{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:45P2BA3YOADWNXDUKK2A6HGFMA","short_pith_number":"pith:45P2BA3Y","schema_version":"1.0","canonical_sha256":"e75fa08378700766dc7452b40f1cc5601ab5ad6eb41c96579dbcabee6ad2a08d","source":{"kind":"arxiv","id":"2410.12972","version":3},"attestation_state":"computed","paper":{"title":"KCIF: Knowledge-Conditioned Instruction Following","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Danish Contractor, Praveen Venkateswaran, Prince Kumar, Rudra Murthy","submitted_at":"2024-10-16T19:07:37Z","abstract_excerpt":"LLM evaluation benchmarks have traditionally separated the testing of knowledge/reasoning capabilities from instruction following. In this work, we study the interaction between knowledge and instruction following, and observe that LLMs struggle to follow simple answer modifying instructions, and are also distracted by instructions that should have no bearing on the original knowledge task answer. We leverage existing multiple-choice answer based knowledge benchmarks and apply a set of simple instructions which include manipulating text (eg.: change case), numeric quantities (eg.: increase val"},"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":"2410.12972","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2024-10-16T19:07:37Z","cross_cats_sorted":[],"title_canon_sha256":"8178385ca73689ad95e3e951de91ad0ff648809440fadfeefa8e2e6b2dbed1c9","abstract_canon_sha256":"4ac9de0a8fc78c550191b3a4ff6539fab500d9991d0772318e2c7acd0d25060d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:08:01.734526Z","signature_b64":"oDot6IYGZuPk616U3tsBfoZdUfBLd9aB88/7rnDzMNt+2IO4CaK79tgssRhQxeYbsEx1a6KS9NMOSFzEY5XLBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e75fa08378700766dc7452b40f1cc5601ab5ad6eb41c96579dbcabee6ad2a08d","last_reissued_at":"2026-07-05T11:08:01.733961Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:08:01.733961Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"KCIF: Knowledge-Conditioned Instruction Following","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Danish Contractor, Praveen Venkateswaran, Prince Kumar, Rudra Murthy","submitted_at":"2024-10-16T19:07:37Z","abstract_excerpt":"LLM evaluation benchmarks have traditionally separated the testing of knowledge/reasoning capabilities from instruction following. In this work, we study the interaction between knowledge and instruction following, and observe that LLMs struggle to follow simple answer modifying instructions, and are also distracted by instructions that should have no bearing on the original knowledge task answer. We leverage existing multiple-choice answer based knowledge benchmarks and apply a set of simple instructions which include manipulating text (eg.: change case), numeric quantities (eg.: increase val"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.12972","kind":"arxiv","version":3},"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/2410.12972/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":"2410.12972","created_at":"2026-07-05T11:08:01.734019+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.12972v3","created_at":"2026-07-05T11:08:01.734019+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.12972","created_at":"2026-07-05T11:08:01.734019+00:00"},{"alias_kind":"pith_short_12","alias_value":"45P2BA3YOADW","created_at":"2026-07-05T11:08:01.734019+00:00"},{"alias_kind":"pith_short_16","alias_value":"45P2BA3YOADWNXDU","created_at":"2026-07-05T11:08:01.734019+00:00"},{"alias_kind":"pith_short_8","alias_value":"45P2BA3Y","created_at":"2026-07-05T11:08:01.734019+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.03465","citing_title":"FINER-SQL: Boosting Small Language Models for Text-to-SQL","ref_index":67,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/45P2BA3YOADWNXDUKK2A6HGFMA","json":"https://pith.science/pith/45P2BA3YOADWNXDUKK2A6HGFMA.json","graph_json":"https://pith.science/api/pith-number/45P2BA3YOADWNXDUKK2A6HGFMA/graph.json","events_json":"https://pith.science/api/pith-number/45P2BA3YOADWNXDUKK2A6HGFMA/events.json","paper":"https://pith.science/paper/45P2BA3Y"},"agent_actions":{"view_html":"https://pith.science/pith/45P2BA3YOADWNXDUKK2A6HGFMA","download_json":"https://pith.science/pith/45P2BA3YOADWNXDUKK2A6HGFMA.json","view_paper":"https://pith.science/paper/45P2BA3Y","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.12972&json=true","fetch_graph":"https://pith.science/api/pith-number/45P2BA3YOADWNXDUKK2A6HGFMA/graph.json","fetch_events":"https://pith.science/api/pith-number/45P2BA3YOADWNXDUKK2A6HGFMA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/45P2BA3YOADWNXDUKK2A6HGFMA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/45P2BA3YOADWNXDUKK2A6HGFMA/action/storage_attestation","attest_author":"https://pith.science/pith/45P2BA3YOADWNXDUKK2A6HGFMA/action/author_attestation","sign_citation":"https://pith.science/pith/45P2BA3YOADWNXDUKK2A6HGFMA/action/citation_signature","submit_replication":"https://pith.science/pith/45P2BA3YOADWNXDUKK2A6HGFMA/action/replication_record"}},"created_at":"2026-07-05T11:08:01.734019+00:00","updated_at":"2026-07-05T11:08:01.734019+00:00"}