{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:33TSJI4E7YHMQ3K7ZUSCLKSSZI","short_pith_number":"pith:33TSJI4E","schema_version":"1.0","canonical_sha256":"dee724a384fe0ec86d5fcd2425aa52ca05a9ac0980ced0d63336ac7d6555a7b0","source":{"kind":"arxiv","id":"2505.19440","version":1},"attestation_state":"computed","paper":{"title":"The Birth of Knowledge: Emergent Features across Time, Space, and Scale in Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Micah Adler, Nir Shavit, Shashata Sawmya","submitted_at":"2025-05-26T02:59:54Z","abstract_excerpt":"This paper studies the emergence of interpretable categorical features within large language models (LLMs), analyzing their behavior across training checkpoints (time), transformer layers (space), and varying model sizes (scale). Using sparse autoencoders for mechanistic interpretability, we identify when and where specific semantic concepts emerge within neural activations. Results indicate clear temporal and scale-specific thresholds for feature emergence across multiple domains. Notably, spatial analysis reveals unexpected semantic reactivation, with early-layer features re-emerging at late"},"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":"2505.19440","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-05-26T02:59:54Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"4a6aa21828d58414fffe05f65be3eebc60a24df519bf140d82567eef763b2d05","abstract_canon_sha256":"3c4e78ef1325561283522ff359c419cb615b12466229be24d520ead02a726a12"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:09:30.307241Z","signature_b64":"BQxO6Xrq26iRR9iMKrQlXeI0hcgOToG4Rgx2l91qVCDRgPdyisSFT5wumTw3HSs3c+il1eafl5rCb938uFt8Aw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"dee724a384fe0ec86d5fcd2425aa52ca05a9ac0980ced0d63336ac7d6555a7b0","last_reissued_at":"2026-07-05T11:09:30.306736Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:09:30.306736Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The Birth of Knowledge: Emergent Features across Time, Space, and Scale in Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Micah Adler, Nir Shavit, Shashata Sawmya","submitted_at":"2025-05-26T02:59:54Z","abstract_excerpt":"This paper studies the emergence of interpretable categorical features within large language models (LLMs), analyzing their behavior across training checkpoints (time), transformer layers (space), and varying model sizes (scale). Using sparse autoencoders for mechanistic interpretability, we identify when and where specific semantic concepts emerge within neural activations. Results indicate clear temporal and scale-specific thresholds for feature emergence across multiple domains. Notably, spatial analysis reveals unexpected semantic reactivation, with early-layer features re-emerging at late"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.19440","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/2505.19440/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":"2505.19440","created_at":"2026-07-05T11:09:30.306795+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.19440v1","created_at":"2026-07-05T11:09:30.306795+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.19440","created_at":"2026-07-05T11:09:30.306795+00:00"},{"alias_kind":"pith_short_12","alias_value":"33TSJI4E7YHM","created_at":"2026-07-05T11:09:30.306795+00:00"},{"alias_kind":"pith_short_16","alias_value":"33TSJI4E7YHMQ3K7","created_at":"2026-07-05T11:09:30.306795+00:00"},{"alias_kind":"pith_short_8","alias_value":"33TSJI4E","created_at":"2026-07-05T11:09:30.306795+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2510.26745","citing_title":"Deep sequence models tend to memorize geometrically; it is unclear why","ref_index":157,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/33TSJI4E7YHMQ3K7ZUSCLKSSZI","json":"https://pith.science/pith/33TSJI4E7YHMQ3K7ZUSCLKSSZI.json","graph_json":"https://pith.science/api/pith-number/33TSJI4E7YHMQ3K7ZUSCLKSSZI/graph.json","events_json":"https://pith.science/api/pith-number/33TSJI4E7YHMQ3K7ZUSCLKSSZI/events.json","paper":"https://pith.science/paper/33TSJI4E"},"agent_actions":{"view_html":"https://pith.science/pith/33TSJI4E7YHMQ3K7ZUSCLKSSZI","download_json":"https://pith.science/pith/33TSJI4E7YHMQ3K7ZUSCLKSSZI.json","view_paper":"https://pith.science/paper/33TSJI4E","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.19440&json=true","fetch_graph":"https://pith.science/api/pith-number/33TSJI4E7YHMQ3K7ZUSCLKSSZI/graph.json","fetch_events":"https://pith.science/api/pith-number/33TSJI4E7YHMQ3K7ZUSCLKSSZI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/33TSJI4E7YHMQ3K7ZUSCLKSSZI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/33TSJI4E7YHMQ3K7ZUSCLKSSZI/action/storage_attestation","attest_author":"https://pith.science/pith/33TSJI4E7YHMQ3K7ZUSCLKSSZI/action/author_attestation","sign_citation":"https://pith.science/pith/33TSJI4E7YHMQ3K7ZUSCLKSSZI/action/citation_signature","submit_replication":"https://pith.science/pith/33TSJI4E7YHMQ3K7ZUSCLKSSZI/action/replication_record"}},"created_at":"2026-07-05T11:09:30.306795+00:00","updated_at":"2026-07-05T11:09:30.306795+00:00"}