{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:3IGB7LQSUWTZ7RYQZS6PBYOHC3","short_pith_number":"pith:3IGB7LQS","schema_version":"1.0","canonical_sha256":"da0c1fae12a5a79fc710ccbcf0e1c716c9b56a29fc71f1ecec6d92d62a7642d9","source":{"kind":"arxiv","id":"2410.01444","version":5},"attestation_state":"computed","paper":{"title":"Geometric Signatures of Compositionality Across a Language Model's Lifetime","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.IT","cs.LG","math.IT"],"primary_cat":"cs.CL","authors_text":"Emily Cheng, Jin Hwa Lee, Lei Yu, Thomas Jiralerspong, Yoshua Bengio","submitted_at":"2024-10-02T11:54:06Z","abstract_excerpt":"By virtue of linguistic compositionality, few syntactic rules and a finite lexicon can generate an unbounded number of sentences. That is, language, though seemingly high-dimensional, can be explained using relatively few degrees of freedom. An open question is whether contemporary language models (LMs) reflect the intrinsic simplicity of language that is enabled by compositionality. We take a geometric view of this problem by relating the degree of compositionality in a dataset to the intrinsic dimension (ID) of its representations under an LM, a measure of feature complexity. We find not onl"},"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":"2410.01444","kind":"arxiv","version":5},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-10-02T11:54:06Z","cross_cats_sorted":["cs.AI","cs.IT","cs.LG","math.IT"],"title_canon_sha256":"b70e22aeace115062663f97878e14f530fe161472bed7703aae6098b8b8bed4b","abstract_canon_sha256":"ec74f43c454a8450ae2d6cd3c0c39fb45fe9a35003ab6147bf75e0a46060e032"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:22:32.597687Z","signature_b64":"zXFcNS+u19SmrfbSL74d6faWmZ/3mfuGPyiVOOJmbDCO27FhV3MEx6h9q44rdu1CEdvbJNyLPqctkNEviWJkCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"da0c1fae12a5a79fc710ccbcf0e1c716c9b56a29fc71f1ecec6d92d62a7642d9","last_reissued_at":"2026-07-05T11:22:32.597205Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:22:32.597205Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Geometric Signatures of Compositionality Across a Language Model's Lifetime","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.IT","cs.LG","math.IT"],"primary_cat":"cs.CL","authors_text":"Emily Cheng, Jin Hwa Lee, Lei Yu, Thomas Jiralerspong, Yoshua Bengio","submitted_at":"2024-10-02T11:54:06Z","abstract_excerpt":"By virtue of linguistic compositionality, few syntactic rules and a finite lexicon can generate an unbounded number of sentences. That is, language, though seemingly high-dimensional, can be explained using relatively few degrees of freedom. An open question is whether contemporary language models (LMs) reflect the intrinsic simplicity of language that is enabled by compositionality. We take a geometric view of this problem by relating the degree of compositionality in a dataset to the intrinsic dimension (ID) of its representations under an LM, a measure of feature complexity. We find not onl"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.01444","kind":"arxiv","version":5},"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/2410.01444/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":"2410.01444","created_at":"2026-07-05T11:22:32.597268+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.01444v5","created_at":"2026-07-05T11:22:32.597268+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.01444","created_at":"2026-07-05T11:22:32.597268+00:00"},{"alias_kind":"pith_short_12","alias_value":"3IGB7LQSUWTZ","created_at":"2026-07-05T11:22:32.597268+00:00"},{"alias_kind":"pith_short_16","alias_value":"3IGB7LQSUWTZ7RYQ","created_at":"2026-07-05T11:22:32.597268+00:00"},{"alias_kind":"pith_short_8","alias_value":"3IGB7LQS","created_at":"2026-07-05T11:22:32.597268+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.05194","citing_title":"Temporal Preference Concepts and their Functions in a Large Language Model","ref_index":64,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01381","citing_title":"A framework for analyzing concept representations in neural models","ref_index":223,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/3IGB7LQSUWTZ7RYQZS6PBYOHC3","json":"https://pith.science/pith/3IGB7LQSUWTZ7RYQZS6PBYOHC3.json","graph_json":"https://pith.science/api/pith-number/3IGB7LQSUWTZ7RYQZS6PBYOHC3/graph.json","events_json":"https://pith.science/api/pith-number/3IGB7LQSUWTZ7RYQZS6PBYOHC3/events.json","paper":"https://pith.science/paper/3IGB7LQS"},"agent_actions":{"view_html":"https://pith.science/pith/3IGB7LQSUWTZ7RYQZS6PBYOHC3","download_json":"https://pith.science/pith/3IGB7LQSUWTZ7RYQZS6PBYOHC3.json","view_paper":"https://pith.science/paper/3IGB7LQS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.01444&json=true","fetch_graph":"https://pith.science/api/pith-number/3IGB7LQSUWTZ7RYQZS6PBYOHC3/graph.json","fetch_events":"https://pith.science/api/pith-number/3IGB7LQSUWTZ7RYQZS6PBYOHC3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/3IGB7LQSUWTZ7RYQZS6PBYOHC3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/3IGB7LQSUWTZ7RYQZS6PBYOHC3/action/storage_attestation","attest_author":"https://pith.science/pith/3IGB7LQSUWTZ7RYQZS6PBYOHC3/action/author_attestation","sign_citation":"https://pith.science/pith/3IGB7LQSUWTZ7RYQZS6PBYOHC3/action/citation_signature","submit_replication":"https://pith.science/pith/3IGB7LQSUWTZ7RYQZS6PBYOHC3/action/replication_record"}},"created_at":"2026-07-05T11:22:32.597268+00:00","updated_at":"2026-07-05T11:22:32.597268+00:00"}