{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:V62H6WBL72GKNUY4R2JUDUYFJS","short_pith_number":"pith:V62H6WBL","schema_version":"1.0","canonical_sha256":"afb47f582bfe8ca6d31c8e9341d3054cb93957ee68fec34520b952408fe1d48f","source":{"kind":"arxiv","id":"2504.08775","version":3},"attestation_state":"computed","paper":{"title":"Layers at Similar Depths Generate Similar Activations Across LLM Architectures","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Aaron Schein, Christopher Wolfram","submitted_at":"2025-04-03T21:02:30Z","abstract_excerpt":"How do the latent spaces used by independently-trained LLMs relate to one another? We study the nearest neighbor relationships induced by activations at different layers of 24 open-weight LLMs, and find that they 1) tend to vary from layer to layer within a model, and 2) are approximately shared between corresponding layers of different models. Claim 2 shows that these nearest neighbor relationships are not arbitrary, as they are shared across models, but Claim 1 shows that they are not \"obvious\" either, as there is no single set of nearest neighbor relationships that is universally shared. To"},"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":"2504.08775","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-04-03T21:02:30Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"dc7b05ef3a4e12db7d5d18ff183622618ef492f921653ee0892dc063673dc96c","abstract_canon_sha256":"b477faf8cbc9b4c2387055b3bc7c1629aac787ce9aa379d0305605e1f2942e68"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:50:34.766479Z","signature_b64":"aCh+wNj72FB0yG+kRpqHEvjBxvacWBegoDcM1I3vlLsHpwUsssD808VwtrryR+tkcOvxzw9KHj+gme3uWwcSDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"afb47f582bfe8ca6d31c8e9341d3054cb93957ee68fec34520b952408fe1d48f","last_reissued_at":"2026-07-05T11:50:34.766038Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:50:34.766038Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Layers at Similar Depths Generate Similar Activations Across LLM Architectures","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Aaron Schein, Christopher Wolfram","submitted_at":"2025-04-03T21:02:30Z","abstract_excerpt":"How do the latent spaces used by independently-trained LLMs relate to one another? We study the nearest neighbor relationships induced by activations at different layers of 24 open-weight LLMs, and find that they 1) tend to vary from layer to layer within a model, and 2) are approximately shared between corresponding layers of different models. Claim 2 shows that these nearest neighbor relationships are not arbitrary, as they are shared across models, but Claim 1 shows that they are not \"obvious\" either, as there is no single set of nearest neighbor relationships that is universally shared. To"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.08775","kind":"arxiv","version":3},"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/2504.08775/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":"2504.08775","created_at":"2026-07-05T11:50:34.766096+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.08775v3","created_at":"2026-07-05T11:50:34.766096+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.08775","created_at":"2026-07-05T11:50:34.766096+00:00"},{"alias_kind":"pith_short_12","alias_value":"V62H6WBL72GK","created_at":"2026-07-05T11:50:34.766096+00:00"},{"alias_kind":"pith_short_16","alias_value":"V62H6WBL72GKNUY4","created_at":"2026-07-05T11:50:34.766096+00:00"},{"alias_kind":"pith_short_8","alias_value":"V62H6WBL","created_at":"2026-07-05T11:50:34.766096+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.08561","citing_title":"Contravariance Theory: Strong Alignment for Minimal Solutions to Hard Tasks","ref_index":2,"is_internal_anchor":true},{"citing_arxiv_id":"2606.30813","citing_title":"Gradient Smoothing: Coupling Layer-wise Updates for Improved Optimization","ref_index":36,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/V62H6WBL72GKNUY4R2JUDUYFJS","json":"https://pith.science/pith/V62H6WBL72GKNUY4R2JUDUYFJS.json","graph_json":"https://pith.science/api/pith-number/V62H6WBL72GKNUY4R2JUDUYFJS/graph.json","events_json":"https://pith.science/api/pith-number/V62H6WBL72GKNUY4R2JUDUYFJS/events.json","paper":"https://pith.science/paper/V62H6WBL"},"agent_actions":{"view_html":"https://pith.science/pith/V62H6WBL72GKNUY4R2JUDUYFJS","download_json":"https://pith.science/pith/V62H6WBL72GKNUY4R2JUDUYFJS.json","view_paper":"https://pith.science/paper/V62H6WBL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.08775&json=true","fetch_graph":"https://pith.science/api/pith-number/V62H6WBL72GKNUY4R2JUDUYFJS/graph.json","fetch_events":"https://pith.science/api/pith-number/V62H6WBL72GKNUY4R2JUDUYFJS/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/V62H6WBL72GKNUY4R2JUDUYFJS/action/timestamp_anchor","attest_storage":"https://pith.science/pith/V62H6WBL72GKNUY4R2JUDUYFJS/action/storage_attestation","attest_author":"https://pith.science/pith/V62H6WBL72GKNUY4R2JUDUYFJS/action/author_attestation","sign_citation":"https://pith.science/pith/V62H6WBL72GKNUY4R2JUDUYFJS/action/citation_signature","submit_replication":"https://pith.science/pith/V62H6WBL72GKNUY4R2JUDUYFJS/action/replication_record"}},"created_at":"2026-07-05T11:50:34.766096+00:00","updated_at":"2026-07-05T11:50:34.766096+00:00"}