{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:BRC5VHGN4KZ2E7RPNFNPOB3S5C","short_pith_number":"pith:BRC5VHGN","schema_version":"1.0","canonical_sha256":"0c45da9ccde2b3a27e2f695af70772e88bf2246ea1eccb04e2462d2f64a13a69","source":{"kind":"arxiv","id":"2607.19618","version":1},"attestation_state":"computed","paper":{"title":"Causal dictionary learning reveals and validates transcription-factor binding features in genomic language models","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"q-bio.GN","authors_text":"Sarwan Ali","submitted_at":"2026-07-21T22:54:46Z","abstract_excerpt":"Genomic language models achieve strong performance across regulatory-genomics tasks, yet what these models internally represent remains opaque, and the field lacks a principled procedure for verifying that an apparent ``concept'' inside a model is real rather than an artifact of sequence composition. We introduce a framework that combines sparse dictionary learning with causal intervention to extract, validate, and causally test interpretable features in genomic foundation models. Training top-$k$ sparse autoencoders on the hidden activations of two architecturally distinct models, Nucleotide "},"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":"2607.19618","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"q-bio.GN","submitted_at":"2026-07-21T22:54:46Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"de779ec0929fbd31ae8c3f2a062c4038b710c31496c4130a27d7c30b8e06f58f","abstract_canon_sha256":"bc5e7790b217736af6a5aa453bb772ebc70a913dec0bd42d790ac4ae290fba3e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-23T00:23:59.550199Z","signature_b64":"aoWXwr31qg6FmIUXqqYHSpNchDt/DnXiVWs/kfZ9n0ZYe9Odm5FuvtptTtMj/j4X9UsH6CD60B9T3BnikZXyDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0c45da9ccde2b3a27e2f695af70772e88bf2246ea1eccb04e2462d2f64a13a69","last_reissued_at":"2026-07-23T00:23:59.549303Z","signature_status":"signed_v1","first_computed_at":"2026-07-23T00:23:59.549303Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Causal dictionary learning reveals and validates transcription-factor binding features in genomic language models","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"q-bio.GN","authors_text":"Sarwan Ali","submitted_at":"2026-07-21T22:54:46Z","abstract_excerpt":"Genomic language models achieve strong performance across regulatory-genomics tasks, yet what these models internally represent remains opaque, and the field lacks a principled procedure for verifying that an apparent ``concept'' inside a model is real rather than an artifact of sequence composition. We introduce a framework that combines sparse dictionary learning with causal intervention to extract, validate, and causally test interpretable features in genomic foundation models. Training top-$k$ sparse autoencoders on the hidden activations of two architecturally distinct models, Nucleotide "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.19618","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/2607.19618/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":"2607.19618","created_at":"2026-07-23T00:23:59.549760+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.19618v1","created_at":"2026-07-23T00:23:59.549760+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.19618","created_at":"2026-07-23T00:23:59.549760+00:00"},{"alias_kind":"pith_short_12","alias_value":"BRC5VHGN4KZ2","created_at":"2026-07-23T00:23:59.549760+00:00"},{"alias_kind":"pith_short_16","alias_value":"BRC5VHGN4KZ2E7RP","created_at":"2026-07-23T00:23:59.549760+00:00"},{"alias_kind":"pith_short_8","alias_value":"BRC5VHGN","created_at":"2026-07-23T00:23:59.549760+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/BRC5VHGN4KZ2E7RPNFNPOB3S5C","json":"https://pith.science/pith/BRC5VHGN4KZ2E7RPNFNPOB3S5C.json","graph_json":"https://pith.science/api/pith-number/BRC5VHGN4KZ2E7RPNFNPOB3S5C/graph.json","events_json":"https://pith.science/api/pith-number/BRC5VHGN4KZ2E7RPNFNPOB3S5C/events.json","paper":"https://pith.science/paper/BRC5VHGN"},"agent_actions":{"view_html":"https://pith.science/pith/BRC5VHGN4KZ2E7RPNFNPOB3S5C","download_json":"https://pith.science/pith/BRC5VHGN4KZ2E7RPNFNPOB3S5C.json","view_paper":"https://pith.science/paper/BRC5VHGN","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.19618&json=true","fetch_graph":"https://pith.science/api/pith-number/BRC5VHGN4KZ2E7RPNFNPOB3S5C/graph.json","fetch_events":"https://pith.science/api/pith-number/BRC5VHGN4KZ2E7RPNFNPOB3S5C/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BRC5VHGN4KZ2E7RPNFNPOB3S5C/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BRC5VHGN4KZ2E7RPNFNPOB3S5C/action/storage_attestation","attest_author":"https://pith.science/pith/BRC5VHGN4KZ2E7RPNFNPOB3S5C/action/author_attestation","sign_citation":"https://pith.science/pith/BRC5VHGN4KZ2E7RPNFNPOB3S5C/action/citation_signature","submit_replication":"https://pith.science/pith/BRC5VHGN4KZ2E7RPNFNPOB3S5C/action/replication_record"}},"created_at":"2026-07-23T00:23:59.549760+00:00","updated_at":"2026-07-23T00:23:59.549760+00:00"}