{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:KG333TX2JQTHUCJNCH2JNIKQXB","short_pith_number":"pith:KG333TX2","schema_version":"1.0","canonical_sha256":"51b7bdcefa4c267a092d11f496a150b863820be47073989256fabf5227553940","source":{"kind":"arxiv","id":"2303.02536","version":4},"attestation_state":"computed","paper":{"title":"Finding Alignments Between Interpretable Causal Variables and Distributed Neural Representations","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Atticus Geiger, Christopher Potts, Noah D. Goodman, Thomas Icard, Zhengxuan Wu","submitted_at":"2023-03-05T00:57:49Z","abstract_excerpt":"Causal abstraction is a promising theoretical framework for explainable artificial intelligence that defines when an interpretable high-level causal model is a faithful simplification of a low-level deep learning system. However, existing causal abstraction methods have two major limitations: they require a brute-force search over alignments between the high-level model and the low-level one, and they presuppose that variables in the high-level model will align with disjoint sets of neurons in the low-level one. In this paper, we present distributed alignment search (DAS), which overcomes thes"},"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":"2303.02536","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2023-03-05T00:57:49Z","cross_cats_sorted":[],"title_canon_sha256":"0975d925ed2c1a2cf90fdf022daa63d978cfd38ba66f0298d7fcdabe31b1deac","abstract_canon_sha256":"04d423556d64f099693378f01d3d555c4448d64a9478f7ca29f149da58b3380e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:47:55.427840Z","signature_b64":"5ikxG642MwP0ZvD6pgFFfB3zzKWn7J6a94X2HYK/fkWAlqccixdOzlC/C74zjwDza0Y4tLI8Xvaae1HMd4e4Aw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"51b7bdcefa4c267a092d11f496a150b863820be47073989256fabf5227553940","last_reissued_at":"2026-07-05T07:47:55.427228Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:47:55.427228Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Finding Alignments Between Interpretable Causal Variables and Distributed Neural Representations","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Atticus Geiger, Christopher Potts, Noah D. Goodman, Thomas Icard, Zhengxuan Wu","submitted_at":"2023-03-05T00:57:49Z","abstract_excerpt":"Causal abstraction is a promising theoretical framework for explainable artificial intelligence that defines when an interpretable high-level causal model is a faithful simplification of a low-level deep learning system. However, existing causal abstraction methods have two major limitations: they require a brute-force search over alignments between the high-level model and the low-level one, and they presuppose that variables in the high-level model will align with disjoint sets of neurons in the low-level one. In this paper, we present distributed alignment search (DAS), which overcomes thes"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2303.02536","kind":"arxiv","version":4},"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/2303.02536/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":"2303.02536","created_at":"2026-07-05T07:47:55.427338+00:00"},{"alias_kind":"arxiv_version","alias_value":"2303.02536v4","created_at":"2026-07-05T07:47:55.427338+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2303.02536","created_at":"2026-07-05T07:47:55.427338+00:00"},{"alias_kind":"pith_short_12","alias_value":"KG333TX2JQTH","created_at":"2026-07-05T07:47:55.427338+00:00"},{"alias_kind":"pith_short_16","alias_value":"KG333TX2JQTHUCJN","created_at":"2026-07-05T07:47:55.427338+00:00"},{"alias_kind":"pith_short_8","alias_value":"KG333TX2","created_at":"2026-07-05T07:47:55.427338+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":7,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2309.16042","citing_title":"Towards Best Practices of Activation Patching in Language Models: Metrics and Methods","ref_index":80,"is_internal_anchor":false},{"citing_arxiv_id":"2304.05969","citing_title":"Localizing Model Behavior with Path Patching","ref_index":43,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12809","citing_title":"Correcting Influence: Unboxing LLM Outputs with Orthogonal Latent Spaces","ref_index":75,"is_internal_anchor":false},{"citing_arxiv_id":"2604.27401","citing_title":"Perturbation Probing: A Two-Pass-per-Prompt Diagnostic for FFN Behavioral Circuits in Aligned LLMs","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08934","citing_title":"From Mechanistic to Compositional Interpretability","ref_index":148,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05715","citing_title":"Decodable but Not Corrected by Fixed Residual-Stream Linear Steering: Evidence from Medical LLM Failure Regimes","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2604.19018","citing_title":"Local Linearity of LLMs Enables Activation Steering via Model-Based Linear Optimal Control","ref_index":46,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KG333TX2JQTHUCJNCH2JNIKQXB","json":"https://pith.science/pith/KG333TX2JQTHUCJNCH2JNIKQXB.json","graph_json":"https://pith.science/api/pith-number/KG333TX2JQTHUCJNCH2JNIKQXB/graph.json","events_json":"https://pith.science/api/pith-number/KG333TX2JQTHUCJNCH2JNIKQXB/events.json","paper":"https://pith.science/paper/KG333TX2"},"agent_actions":{"view_html":"https://pith.science/pith/KG333TX2JQTHUCJNCH2JNIKQXB","download_json":"https://pith.science/pith/KG333TX2JQTHUCJNCH2JNIKQXB.json","view_paper":"https://pith.science/paper/KG333TX2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2303.02536&json=true","fetch_graph":"https://pith.science/api/pith-number/KG333TX2JQTHUCJNCH2JNIKQXB/graph.json","fetch_events":"https://pith.science/api/pith-number/KG333TX2JQTHUCJNCH2JNIKQXB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KG333TX2JQTHUCJNCH2JNIKQXB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KG333TX2JQTHUCJNCH2JNIKQXB/action/storage_attestation","attest_author":"https://pith.science/pith/KG333TX2JQTHUCJNCH2JNIKQXB/action/author_attestation","sign_citation":"https://pith.science/pith/KG333TX2JQTHUCJNCH2JNIKQXB/action/citation_signature","submit_replication":"https://pith.science/pith/KG333TX2JQTHUCJNCH2JNIKQXB/action/replication_record"}},"created_at":"2026-07-05T07:47:55.427338+00:00","updated_at":"2026-07-05T07:47:55.427338+00:00"}