{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:T2WLR2RBRBJYWHO55JDR5E55N2","short_pith_number":"pith:T2WLR2RB","canonical_record":{"source":{"id":"2406.14785","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-20T23:27:06Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"ec624b198c39e2dfb46f4fb898162bc1fed97ab40bc75f79c299ab244cd891f5","abstract_canon_sha256":"8c2ef70863692aae3b9dd1ff1c9e61985b9361146c24a04cbc2f2b7d84a04a4e"},"schema_version":"1.0"},"canonical_sha256":"9eacb8ea2188538b1dddea471e93bd6e83de4768aa22d071733f71f7b47c66ae","source":{"kind":"arxiv","id":"2406.14785","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2406.14785","created_at":"2026-07-05T08:35:08Z"},{"alias_kind":"arxiv_version","alias_value":"2406.14785v1","created_at":"2026-07-05T08:35:08Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.14785","created_at":"2026-07-05T08:35:08Z"},{"alias_kind":"pith_short_12","alias_value":"T2WLR2RBRBJY","created_at":"2026-07-05T08:35:08Z"},{"alias_kind":"pith_short_16","alias_value":"T2WLR2RBRBJYWHO5","created_at":"2026-07-05T08:35:08Z"},{"alias_kind":"pith_short_8","alias_value":"T2WLR2RB","created_at":"2026-07-05T08:35:08Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:T2WLR2RBRBJYWHO55JDR5E55N2","target":"record","payload":{"canonical_record":{"source":{"id":"2406.14785","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-20T23:27:06Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"ec624b198c39e2dfb46f4fb898162bc1fed97ab40bc75f79c299ab244cd891f5","abstract_canon_sha256":"8c2ef70863692aae3b9dd1ff1c9e61985b9361146c24a04cbc2f2b7d84a04a4e"},"schema_version":"1.0"},"canonical_sha256":"9eacb8ea2188538b1dddea471e93bd6e83de4768aa22d071733f71f7b47c66ae","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:35:08.383475Z","signature_b64":"hmglRJcY8kjF97d7D3iCC7WTpa1ZUexnnlvaAjYpii7c5Zojj+/JIq2i318SROBUlkZeuCAA/4yjRQP6Ozb7DQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9eacb8ea2188538b1dddea471e93bd6e83de4768aa22d071733f71f7b47c66ae","last_reissued_at":"2026-07-05T08:35:08.382934Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:35:08.382934Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2406.14785","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T08:35:08Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"mkCtRib956hIPNoTyxoH9fDetyKZ3eoouOgybenDMaBypeGBMVaw+/88YTm8wjUJUpsGwyrSb1zZED5oaRrJAA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T07:19:42.367574Z"},"content_sha256":"33dfd5f09054dfc85597b5ff6f2ee5dd37e7b2a18aaf5e121c98a52940cf7104","schema_version":"1.0","event_id":"sha256:33dfd5f09054dfc85597b5ff6f2ee5dd37e7b2a18aaf5e121c98a52940cf7104"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:T2WLR2RBRBJYWHO55JDR5E55N2","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Understanding Finetuning for Factual Knowledge Extraction","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Aditi Raghunathan, Gaurav Ghosal, Tatsunori Hashimoto","submitted_at":"2024-06-20T23:27:06Z","abstract_excerpt":"In this work, we study the impact of QA fine-tuning data on downstream factuality. We show that fine-tuning on lesser-known facts that are poorly stored during pretraining yields significantly worse factuality than fine-tuning on well-known facts, even when all facts are seen during pretraining. We prove this phenomenon theoretically, showing that training on lesser-known facts can lead the model to ignore subject entity names and instead output a generic plausible response even when the relevant factual knowledge is encoded in the model. On three question answering benchmarks (PopQA, Entity Q"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.14785","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/2406.14785/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T08:35:08Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"wfYGPik3ToClghECnGt788j/8f4tEj1jkUHfGcKRRkRF+qsL0S745U4mQqrWmcLkWiDwtKEG8anK8WWLGbmHCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T07:19:42.367996Z"},"content_sha256":"cb697bd543f1ff5ace58ade9e7ee92170c2f544d04f4eb97e159fe4c479926d3","schema_version":"1.0","event_id":"sha256:cb697bd543f1ff5ace58ade9e7ee92170c2f544d04f4eb97e159fe4c479926d3"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/T2WLR2RBRBJYWHO55JDR5E55N2/bundle.json","state_url":"https://pith.science/pith/T2WLR2RBRBJYWHO55JDR5E55N2/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/T2WLR2RBRBJYWHO55JDR5E55N2/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-06T07:19:42Z","links":{"resolver":"https://pith.science/pith/T2WLR2RBRBJYWHO55JDR5E55N2","bundle":"https://pith.science/pith/T2WLR2RBRBJYWHO55JDR5E55N2/bundle.json","state":"https://pith.science/pith/T2WLR2RBRBJYWHO55JDR5E55N2/state.json","well_known_bundle":"https://pith.science/.well-known/pith/T2WLR2RBRBJYWHO55JDR5E55N2/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:T2WLR2RBRBJYWHO55JDR5E55N2","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"8c2ef70863692aae3b9dd1ff1c9e61985b9361146c24a04cbc2f2b7d84a04a4e","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-20T23:27:06Z","title_canon_sha256":"ec624b198c39e2dfb46f4fb898162bc1fed97ab40bc75f79c299ab244cd891f5"},"schema_version":"1.0","source":{"id":"2406.14785","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2406.14785","created_at":"2026-07-05T08:35:08Z"},{"alias_kind":"arxiv_version","alias_value":"2406.14785v1","created_at":"2026-07-05T08:35:08Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.14785","created_at":"2026-07-05T08:35:08Z"},{"alias_kind":"pith_short_12","alias_value":"T2WLR2RBRBJY","created_at":"2026-07-05T08:35:08Z"},{"alias_kind":"pith_short_16","alias_value":"T2WLR2RBRBJYWHO5","created_at":"2026-07-05T08:35:08Z"},{"alias_kind":"pith_short_8","alias_value":"T2WLR2RB","created_at":"2026-07-05T08:35:08Z"}],"graph_snapshots":[{"event_id":"sha256:cb697bd543f1ff5ace58ade9e7ee92170c2f544d04f4eb97e159fe4c479926d3","target":"graph","created_at":"2026-07-05T08:35:08Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2406.14785/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"In this work, we study the impact of QA fine-tuning data on downstream factuality. We show that fine-tuning on lesser-known facts that are poorly stored during pretraining yields significantly worse factuality than fine-tuning on well-known facts, even when all facts are seen during pretraining. We prove this phenomenon theoretically, showing that training on lesser-known facts can lead the model to ignore subject entity names and instead output a generic plausible response even when the relevant factual knowledge is encoded in the model. On three question answering benchmarks (PopQA, Entity Q","authors_text":"Aditi Raghunathan, Gaurav Ghosal, Tatsunori Hashimoto","cross_cats":["cs.LG"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-20T23:27:06Z","title":"Understanding Finetuning for Factual Knowledge Extraction"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.14785","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:33dfd5f09054dfc85597b5ff6f2ee5dd37e7b2a18aaf5e121c98a52940cf7104","target":"record","created_at":"2026-07-05T08:35:08Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"8c2ef70863692aae3b9dd1ff1c9e61985b9361146c24a04cbc2f2b7d84a04a4e","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-20T23:27:06Z","title_canon_sha256":"ec624b198c39e2dfb46f4fb898162bc1fed97ab40bc75f79c299ab244cd891f5"},"schema_version":"1.0","source":{"id":"2406.14785","kind":"arxiv","version":1}},"canonical_sha256":"9eacb8ea2188538b1dddea471e93bd6e83de4768aa22d071733f71f7b47c66ae","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"9eacb8ea2188538b1dddea471e93bd6e83de4768aa22d071733f71f7b47c66ae","first_computed_at":"2026-07-05T08:35:08.382934Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T08:35:08.382934Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"hmglRJcY8kjF97d7D3iCC7WTpa1ZUexnnlvaAjYpii7c5Zojj+/JIq2i318SROBUlkZeuCAA/4yjRQP6Ozb7DQ==","signature_status":"signed_v1","signed_at":"2026-07-05T08:35:08.383475Z","signed_message":"canonical_sha256_bytes"},"source_id":"2406.14785","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:33dfd5f09054dfc85597b5ff6f2ee5dd37e7b2a18aaf5e121c98a52940cf7104","sha256:cb697bd543f1ff5ace58ade9e7ee92170c2f544d04f4eb97e159fe4c479926d3"],"state_sha256":"96473bc0d134138c65db823c660801e115caec1775c5671efe9eb26a14f33de9"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Zuzx+STfRlVdzOdweXFeEt5bVC166ITvQNoAARKEEzPB+BmpauPtXyPcJwUqOJphIGQhAyxHyxSuQyZ0foZnCA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-06T07:19:42.372087Z","bundle_sha256":"e4ea453e4ea5764618158056b812e3e2ddc36d691b9d66b8d8518e006056e101"}}