{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:XAEXHRN5IWIDEZT7ROPEGWMUSN","short_pith_number":"pith:XAEXHRN5","schema_version":"1.0","canonical_sha256":"b80973c5bd459032667f8b9e4359949378a48561c716f8558c469ebe59d4baf7","source":{"kind":"arxiv","id":"2412.05571","version":1},"attestation_state":"computed","paper":{"title":"A polar coordinate system represents syntax in large language models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Emmanuel Chemla, Jean-R\\'emi King, Pablo Diego-Sim\\'on, St\\'ephane d'Ascoli, Yair Lakretz","submitted_at":"2024-12-07T07:37:20Z","abstract_excerpt":"Originally formalized with symbolic representations, syntactic trees may also be effectively represented in the activations of large language models (LLMs). Indeed, a 'Structural Probe' can find a subspace of neural activations, where syntactically related words are relatively close to one-another. However, this syntactic code remains incomplete: the distance between the Structural Probe word embeddings can represent the existence but not the type and direction of syntactic relations. Here, we hypothesize that syntactic relations are, in fact, coded by the relative direction between nearby emb"},"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":"2412.05571","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-12-07T07:37:20Z","cross_cats_sorted":[],"title_canon_sha256":"710d71d155606c7c5f1274b392b5301395a0a1c82b8944d3202d1dc1f0d6f837","abstract_canon_sha256":"72c2ca30f2771f00667e992f4aeffd05184d83114b6a096a9b5ce001da5491e4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:46:05.569809Z","signature_b64":"VzWXAyWUgLlrqnOHmF7sl6fU17oyfi4cfZ5kybd7e29csvcYdJwUpeM8Tky0ZW5sA+h4Fxyeg/6VECZ0LiJUAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b80973c5bd459032667f8b9e4359949378a48561c716f8558c469ebe59d4baf7","last_reissued_at":"2026-07-05T09:46:05.569339Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:46:05.569339Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A polar coordinate system represents syntax in large language models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Emmanuel Chemla, Jean-R\\'emi King, Pablo Diego-Sim\\'on, St\\'ephane d'Ascoli, Yair Lakretz","submitted_at":"2024-12-07T07:37:20Z","abstract_excerpt":"Originally formalized with symbolic representations, syntactic trees may also be effectively represented in the activations of large language models (LLMs). Indeed, a 'Structural Probe' can find a subspace of neural activations, where syntactically related words are relatively close to one-another. However, this syntactic code remains incomplete: the distance between the Structural Probe word embeddings can represent the existence but not the type and direction of syntactic relations. Here, we hypothesize that syntactic relations are, in fact, coded by the relative direction between nearby emb"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.05571","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/2412.05571/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":"2412.05571","created_at":"2026-07-05T09:46:05.569398+00:00"},{"alias_kind":"arxiv_version","alias_value":"2412.05571v1","created_at":"2026-07-05T09:46:05.569398+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.05571","created_at":"2026-07-05T09:46:05.569398+00:00"},{"alias_kind":"pith_short_12","alias_value":"XAEXHRN5IWID","created_at":"2026-07-05T09:46:05.569398+00:00"},{"alias_kind":"pith_short_16","alias_value":"XAEXHRN5IWIDEZT7","created_at":"2026-07-05T09:46:05.569398+00:00"},{"alias_kind":"pith_short_8","alias_value":"XAEXHRN5","created_at":"2026-07-05T09:46:05.569398+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.29634","citing_title":"Relational Rank Geometry in Transformers: Detecting and Steering Hidden-State Relation Frames","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2602.00986","citing_title":"Sparse Reward Subsystem in Large Language Models","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01381","citing_title":"A framework for analyzing concept representations in neural models","ref_index":171,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/XAEXHRN5IWIDEZT7ROPEGWMUSN","json":"https://pith.science/pith/XAEXHRN5IWIDEZT7ROPEGWMUSN.json","graph_json":"https://pith.science/api/pith-number/XAEXHRN5IWIDEZT7ROPEGWMUSN/graph.json","events_json":"https://pith.science/api/pith-number/XAEXHRN5IWIDEZT7ROPEGWMUSN/events.json","paper":"https://pith.science/paper/XAEXHRN5"},"agent_actions":{"view_html":"https://pith.science/pith/XAEXHRN5IWIDEZT7ROPEGWMUSN","download_json":"https://pith.science/pith/XAEXHRN5IWIDEZT7ROPEGWMUSN.json","view_paper":"https://pith.science/paper/XAEXHRN5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2412.05571&json=true","fetch_graph":"https://pith.science/api/pith-number/XAEXHRN5IWIDEZT7ROPEGWMUSN/graph.json","fetch_events":"https://pith.science/api/pith-number/XAEXHRN5IWIDEZT7ROPEGWMUSN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XAEXHRN5IWIDEZT7ROPEGWMUSN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XAEXHRN5IWIDEZT7ROPEGWMUSN/action/storage_attestation","attest_author":"https://pith.science/pith/XAEXHRN5IWIDEZT7ROPEGWMUSN/action/author_attestation","sign_citation":"https://pith.science/pith/XAEXHRN5IWIDEZT7ROPEGWMUSN/action/citation_signature","submit_replication":"https://pith.science/pith/XAEXHRN5IWIDEZT7ROPEGWMUSN/action/replication_record"}},"created_at":"2026-07-05T09:46:05.569398+00:00","updated_at":"2026-07-05T09:46:05.569398+00:00"}