{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:YWP32PKSXSETA6TEUGX2QIWKYP","short_pith_number":"pith:YWP32PKS","schema_version":"1.0","canonical_sha256":"c59fbd3d52bc89307a64a1afa822cac3f5c777f506a5f5e2b645005c0d0d21a9","source":{"kind":"arxiv","id":"2210.12574","version":1},"attestation_state":"computed","paper":{"title":"The Curious Case of Absolute Position Embeddings","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Adina Williams, Amirhossein Kazemnejad, Dieuwke Hupkes, Joelle Pineau, Koustuv Sinha, Siva Reddy","submitted_at":"2022-10-23T00:00:04Z","abstract_excerpt":"Transformer language models encode the notion of word order using positional information. Most commonly, this positional information is represented by absolute position embeddings (APEs), that are learned from the pretraining data. However, in natural language, it is not absolute position that matters, but relative position, and the extent to which APEs can capture this type of information has not been investigated. In this work, we observe that models trained with APE over-rely on positional information to the point that they break-down when subjected to sentences with shifted position inform"},"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":"2210.12574","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2022-10-23T00:00:04Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"5ef372d6a130187c2e6eda082822ff033be0a995e696147b645cad608e98909f","abstract_canon_sha256":"6587248dbfc257fe12e687020643e81c42cc8783baf3ff143a3f1956ddd54717"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:09:30.002706Z","signature_b64":"QdY6Ijc0Znt3g2C8ssTkaREzBP7zYyDeTgjOVAaGbDdzkc4bplRm1yTrCI0H9pfiNita9dml/CDdWyj+iQVgAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c59fbd3d52bc89307a64a1afa822cac3f5c777f506a5f5e2b645005c0d0d21a9","last_reissued_at":"2026-07-05T05:09:30.002219Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:09:30.002219Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The Curious Case of Absolute Position Embeddings","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Adina Williams, Amirhossein Kazemnejad, Dieuwke Hupkes, Joelle Pineau, Koustuv Sinha, Siva Reddy","submitted_at":"2022-10-23T00:00:04Z","abstract_excerpt":"Transformer language models encode the notion of word order using positional information. Most commonly, this positional information is represented by absolute position embeddings (APEs), that are learned from the pretraining data. However, in natural language, it is not absolute position that matters, but relative position, and the extent to which APEs can capture this type of information has not been investigated. In this work, we observe that models trained with APE over-rely on positional information to the point that they break-down when subjected to sentences with shifted position inform"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2210.12574","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/2210.12574/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":"2210.12574","created_at":"2026-07-05T05:09:30.002276+00:00"},{"alias_kind":"arxiv_version","alias_value":"2210.12574v1","created_at":"2026-07-05T05:09:30.002276+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2210.12574","created_at":"2026-07-05T05:09:30.002276+00:00"},{"alias_kind":"pith_short_12","alias_value":"YWP32PKSXSET","created_at":"2026-07-05T05:09:30.002276+00:00"},{"alias_kind":"pith_short_16","alias_value":"YWP32PKSXSETA6TE","created_at":"2026-07-05T05:09:30.002276+00:00"},{"alias_kind":"pith_short_8","alias_value":"YWP32PKS","created_at":"2026-07-05T05:09:30.002276+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2503.01835","citing_title":"Primus: Enforcing Attention Usage for 3D Medical Image Segmentation","ref_index":63,"is_internal_anchor":false},{"citing_arxiv_id":"2512.07805","citing_title":"Group Representational Position Encoding","ref_index":21,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YWP32PKSXSETA6TEUGX2QIWKYP","json":"https://pith.science/pith/YWP32PKSXSETA6TEUGX2QIWKYP.json","graph_json":"https://pith.science/api/pith-number/YWP32PKSXSETA6TEUGX2QIWKYP/graph.json","events_json":"https://pith.science/api/pith-number/YWP32PKSXSETA6TEUGX2QIWKYP/events.json","paper":"https://pith.science/paper/YWP32PKS"},"agent_actions":{"view_html":"https://pith.science/pith/YWP32PKSXSETA6TEUGX2QIWKYP","download_json":"https://pith.science/pith/YWP32PKSXSETA6TEUGX2QIWKYP.json","view_paper":"https://pith.science/paper/YWP32PKS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2210.12574&json=true","fetch_graph":"https://pith.science/api/pith-number/YWP32PKSXSETA6TEUGX2QIWKYP/graph.json","fetch_events":"https://pith.science/api/pith-number/YWP32PKSXSETA6TEUGX2QIWKYP/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YWP32PKSXSETA6TEUGX2QIWKYP/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YWP32PKSXSETA6TEUGX2QIWKYP/action/storage_attestation","attest_author":"https://pith.science/pith/YWP32PKSXSETA6TEUGX2QIWKYP/action/author_attestation","sign_citation":"https://pith.science/pith/YWP32PKSXSETA6TEUGX2QIWKYP/action/citation_signature","submit_replication":"https://pith.science/pith/YWP32PKSXSETA6TEUGX2QIWKYP/action/replication_record"}},"created_at":"2026-07-05T05:09:30.002276+00:00","updated_at":"2026-07-05T05:09:30.002276+00:00"}