{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:I3M4YLKDS7YVVJXOP3MPL2ZPEJ","short_pith_number":"pith:I3M4YLKD","canonical_record":{"source":{"id":"2409.14057","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-09-21T08:13:16Z","cross_cats_sorted":[],"title_canon_sha256":"3308dae5aa69adddb73ead15c936f0826309afe6403e1250bb73b1bfdee2221b","abstract_canon_sha256":"e78c9546bab5fd5b4ecfa472387e7dad918cac3537d02b74653823cc319bca8a"},"schema_version":"1.0"},"canonical_sha256":"46d9cc2d4397f15aa6ee7ed8f5eb2f22544c0232fe22b536247c8e13a1cdd22d","source":{"kind":"arxiv","id":"2409.14057","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2409.14057","created_at":"2026-07-05T11:21:35Z"},{"alias_kind":"arxiv_version","alias_value":"2409.14057v2","created_at":"2026-07-05T11:21:35Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2409.14057","created_at":"2026-07-05T11:21:35Z"},{"alias_kind":"pith_short_12","alias_value":"I3M4YLKDS7YV","created_at":"2026-07-05T11:21:35Z"},{"alias_kind":"pith_short_16","alias_value":"I3M4YLKDS7YVVJXO","created_at":"2026-07-05T11:21:35Z"},{"alias_kind":"pith_short_8","alias_value":"I3M4YLKD","created_at":"2026-07-05T11:21:35Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:I3M4YLKDS7YVVJXOP3MPL2ZPEJ","target":"record","payload":{"canonical_record":{"source":{"id":"2409.14057","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-09-21T08:13:16Z","cross_cats_sorted":[],"title_canon_sha256":"3308dae5aa69adddb73ead15c936f0826309afe6403e1250bb73b1bfdee2221b","abstract_canon_sha256":"e78c9546bab5fd5b4ecfa472387e7dad918cac3537d02b74653823cc319bca8a"},"schema_version":"1.0"},"canonical_sha256":"46d9cc2d4397f15aa6ee7ed8f5eb2f22544c0232fe22b536247c8e13a1cdd22d","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:21:35.911522Z","signature_b64":"lQYJAhOTWW/z4ApAWJxmMYWrAUzhhLaxiu7F/i6nTixxZuw4kcj5+PWi7wLG6vjD0VXW8cornxkuBZHfnRZBCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"46d9cc2d4397f15aa6ee7ed8f5eb2f22544c0232fe22b536247c8e13a1cdd22d","last_reissued_at":"2026-07-05T11:21:35.910952Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:21:35.910952Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2409.14057","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T11:21:35Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"y1Eo/I1UAFgQXpZg9Hh51cug5QkvQf0x7wLSq6zWoXIQ7WKIL+uZKednXfZQFCgXLlOWWC5ES3mMsz3091pFCg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-19T06:58:30.746920Z"},"content_sha256":"b9f1fecf17c4d99ddf35356ab96ba397bb0e7029c4e8ab504c1bf3162ad3ebdd","schema_version":"1.0","event_id":"sha256:b9f1fecf17c4d99ddf35356ab96ba397bb0e7029c4e8ab504c1bf3162ad3ebdd"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:I3M4YLKDS7YVVJXOP3MPL2ZPEJ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Co-occurrence is not Factual Association in Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Ji Wu, Miao Li, Xiao Zhang","submitted_at":"2024-09-21T08:13:16Z","abstract_excerpt":"Pretrained language models can encode a large amount of knowledge and utilize it for various reasoning tasks, yet they can still struggle to learn novel factual knowledge effectively from finetuning on limited textual demonstrations. In this work, we show that the reason for this deficiency is that language models are biased to learn word co-occurrence statistics instead of true factual associations. We identify the differences between two forms of knowledge representation in language models: knowledge in the form of co-occurrence statistics is encoded in the middle layers of the transformer m"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2409.14057","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/2409.14057/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":"2026-07-05T11:21:35Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"+sJro9pdFFNACSVNyfkrSNh4Hd/YAs3KRP/kvPM3ingMm3aXXpgAHmYyYMvc3x9HKxXf4TYlame0RsG2jTAFBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-19T06:58:30.747289Z"},"content_sha256":"2d979b88f1a5a906e626355a5aef3179816b4d5b4b1b7f65598cf114b22200d3","schema_version":"1.0","event_id":"sha256:2d979b88f1a5a906e626355a5aef3179816b4d5b4b1b7f65598cf114b22200d3"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/I3M4YLKDS7YVVJXOP3MPL2ZPEJ/bundle.json","state_url":"https://pith.science/pith/I3M4YLKDS7YVVJXOP3MPL2ZPEJ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/I3M4YLKDS7YVVJXOP3MPL2ZPEJ/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":"2026-08-19T06:58:30Z","links":{"resolver":"https://pith.science/pith/I3M4YLKDS7YVVJXOP3MPL2ZPEJ","bundle":"https://pith.science/pith/I3M4YLKDS7YVVJXOP3MPL2ZPEJ/bundle.json","state":"https://pith.science/pith/I3M4YLKDS7YVVJXOP3MPL2ZPEJ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/I3M4YLKDS7YVVJXOP3MPL2ZPEJ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:I3M4YLKDS7YVVJXOP3MPL2ZPEJ","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":"e78c9546bab5fd5b4ecfa472387e7dad918cac3537d02b74653823cc319bca8a","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-09-21T08:13:16Z","title_canon_sha256":"3308dae5aa69adddb73ead15c936f0826309afe6403e1250bb73b1bfdee2221b"},"schema_version":"1.0","source":{"id":"2409.14057","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2409.14057","created_at":"2026-07-05T11:21:35Z"},{"alias_kind":"arxiv_version","alias_value":"2409.14057v2","created_at":"2026-07-05T11:21:35Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2409.14057","created_at":"2026-07-05T11:21:35Z"},{"alias_kind":"pith_short_12","alias_value":"I3M4YLKDS7YV","created_at":"2026-07-05T11:21:35Z"},{"alias_kind":"pith_short_16","alias_value":"I3M4YLKDS7YVVJXO","created_at":"2026-07-05T11:21:35Z"},{"alias_kind":"pith_short_8","alias_value":"I3M4YLKD","created_at":"2026-07-05T11:21:35Z"}],"graph_snapshots":[{"event_id":"sha256:2d979b88f1a5a906e626355a5aef3179816b4d5b4b1b7f65598cf114b22200d3","target":"graph","created_at":"2026-07-05T11:21:35Z","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/2409.14057/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Pretrained language models can encode a large amount of knowledge and utilize it for various reasoning tasks, yet they can still struggle to learn novel factual knowledge effectively from finetuning on limited textual demonstrations. In this work, we show that the reason for this deficiency is that language models are biased to learn word co-occurrence statistics instead of true factual associations. We identify the differences between two forms of knowledge representation in language models: knowledge in the form of co-occurrence statistics is encoded in the middle layers of the transformer m","authors_text":"Ji Wu, Miao Li, Xiao Zhang","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-09-21T08:13:16Z","title":"Co-occurrence is not Factual Association in Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2409.14057","kind":"arxiv","version":2},"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:b9f1fecf17c4d99ddf35356ab96ba397bb0e7029c4e8ab504c1bf3162ad3ebdd","target":"record","created_at":"2026-07-05T11:21:35Z","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":"e78c9546bab5fd5b4ecfa472387e7dad918cac3537d02b74653823cc319bca8a","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-09-21T08:13:16Z","title_canon_sha256":"3308dae5aa69adddb73ead15c936f0826309afe6403e1250bb73b1bfdee2221b"},"schema_version":"1.0","source":{"id":"2409.14057","kind":"arxiv","version":2}},"canonical_sha256":"46d9cc2d4397f15aa6ee7ed8f5eb2f22544c0232fe22b536247c8e13a1cdd22d","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"46d9cc2d4397f15aa6ee7ed8f5eb2f22544c0232fe22b536247c8e13a1cdd22d","first_computed_at":"2026-07-05T11:21:35.910952Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:21:35.910952Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"lQYJAhOTWW/z4ApAWJxmMYWrAUzhhLaxiu7F/i6nTixxZuw4kcj5+PWi7wLG6vjD0VXW8cornxkuBZHfnRZBCA==","signature_status":"signed_v1","signed_at":"2026-07-05T11:21:35.911522Z","signed_message":"canonical_sha256_bytes"},"source_id":"2409.14057","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:b9f1fecf17c4d99ddf35356ab96ba397bb0e7029c4e8ab504c1bf3162ad3ebdd","sha256:2d979b88f1a5a906e626355a5aef3179816b4d5b4b1b7f65598cf114b22200d3"],"state_sha256":"213f537644e3f60720d67e7431655615ff68ac3ca66180f64603c5c5926beae2"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"8HehrHfeyNZ0NQsdsvSEI9zgCLKg8yIB7PpkQkN0I/FU3uUD+V7uREUluXOiW0TuPkOVRAS5Ty6XnxwghiarAA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-19T06:58:30.749847Z","bundle_sha256":"1789861ed133498853d6387b3825243815345c485db4bcdf9a7bc67a90b9337e"}}