{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:DKDD5SBSSGNK6AJD5KED5E6RCM","short_pith_number":"pith:DKDD5SBS","schema_version":"1.0","canonical_sha256":"1a863ec832919aaf0123ea883e93d11339791c045b1e8180cda18d336f8f6424","source":{"kind":"arxiv","id":"2106.05667","version":1},"attestation_state":"computed","paper":{"title":"GraphiT: Encoding Graph Structure in Transformers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Dexiong Chen, Gr\\'egoire Mialon, Julien Mairal, Margot Selosse","submitted_at":"2021-06-10T11:36:22Z","abstract_excerpt":"We show that viewing graphs as sets of node features and incorporating structural and positional information into a transformer architecture is able to outperform representations learned with classical graph neural networks (GNNs). Our model, GraphiT, encodes such information by (i) leveraging relative positional encoding strategies in self-attention scores based on positive definite kernels on graphs, and (ii) enumerating and encoding local sub-structures such as paths of short length. We thoroughly evaluate these two ideas on many classification and regression tasks, demonstrating the effect"},"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":"2106.05667","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2021-06-10T11:36:22Z","cross_cats_sorted":[],"title_canon_sha256":"8afaa239557a0aa030c49e54d478db28a1c7e93ba5a067455eacf8e2bbf8782c","abstract_canon_sha256":"fc2e0fb78a0d63bd18e85cd00cb59681d75b95b054097b5135cffdaee5e68de4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:48:11.951783Z","signature_b64":"bLEfSXPtpPrrtqqD9sWc1Gpwr9Ucd5kawx4iVKVKQ2/fiSBlwy7WMU4fbALk4YHHSsEmEiOuUy9QI/fsAjumAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1a863ec832919aaf0123ea883e93d11339791c045b1e8180cda18d336f8f6424","last_reissued_at":"2026-07-05T02:48:11.951371Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:48:11.951371Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"GraphiT: Encoding Graph Structure in Transformers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Dexiong Chen, Gr\\'egoire Mialon, Julien Mairal, Margot Selosse","submitted_at":"2021-06-10T11:36:22Z","abstract_excerpt":"We show that viewing graphs as sets of node features and incorporating structural and positional information into a transformer architecture is able to outperform representations learned with classical graph neural networks (GNNs). Our model, GraphiT, encodes such information by (i) leveraging relative positional encoding strategies in self-attention scores based on positive definite kernels on graphs, and (ii) enumerating and encoding local sub-structures such as paths of short length. We thoroughly evaluate these two ideas on many classification and regression tasks, demonstrating the effect"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2106.05667","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/2106.05667/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":"2106.05667","created_at":"2026-07-05T02:48:11.951428+00:00"},{"alias_kind":"arxiv_version","alias_value":"2106.05667v1","created_at":"2026-07-05T02:48:11.951428+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2106.05667","created_at":"2026-07-05T02:48:11.951428+00:00"},{"alias_kind":"pith_short_12","alias_value":"DKDD5SBSSGNK","created_at":"2026-07-05T02:48:11.951428+00:00"},{"alias_kind":"pith_short_16","alias_value":"DKDD5SBSSGNK6AJD","created_at":"2026-07-05T02:48:11.951428+00:00"},{"alias_kind":"pith_short_8","alias_value":"DKDD5SBS","created_at":"2026-07-05T02:48:11.951428+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.25293","citing_title":"Communicability-Inspired Positional Encoding (CIPE)","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00757","citing_title":"RADE: Random Add-Drop Edge as a Regularizer","ref_index":40,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/DKDD5SBSSGNK6AJD5KED5E6RCM","json":"https://pith.science/pith/DKDD5SBSSGNK6AJD5KED5E6RCM.json","graph_json":"https://pith.science/api/pith-number/DKDD5SBSSGNK6AJD5KED5E6RCM/graph.json","events_json":"https://pith.science/api/pith-number/DKDD5SBSSGNK6AJD5KED5E6RCM/events.json","paper":"https://pith.science/paper/DKDD5SBS"},"agent_actions":{"view_html":"https://pith.science/pith/DKDD5SBSSGNK6AJD5KED5E6RCM","download_json":"https://pith.science/pith/DKDD5SBSSGNK6AJD5KED5E6RCM.json","view_paper":"https://pith.science/paper/DKDD5SBS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2106.05667&json=true","fetch_graph":"https://pith.science/api/pith-number/DKDD5SBSSGNK6AJD5KED5E6RCM/graph.json","fetch_events":"https://pith.science/api/pith-number/DKDD5SBSSGNK6AJD5KED5E6RCM/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DKDD5SBSSGNK6AJD5KED5E6RCM/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DKDD5SBSSGNK6AJD5KED5E6RCM/action/storage_attestation","attest_author":"https://pith.science/pith/DKDD5SBSSGNK6AJD5KED5E6RCM/action/author_attestation","sign_citation":"https://pith.science/pith/DKDD5SBSSGNK6AJD5KED5E6RCM/action/citation_signature","submit_replication":"https://pith.science/pith/DKDD5SBSSGNK6AJD5KED5E6RCM/action/replication_record"}},"created_at":"2026-07-05T02:48:11.951428+00:00","updated_at":"2026-07-05T02:48:11.951428+00:00"}