{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:HZR6AAVJS2CPUU4W2MBOOPUWBJ","short_pith_number":"pith:HZR6AAVJ","schema_version":"1.0","canonical_sha256":"3e63e002a99684fa5396d302e73e960a6983f48a23d97f7d2e2208f5a0bad5aa","source":{"kind":"arxiv","id":"2501.14926","version":4},"attestation_state":"computed","paper":{"title":"Interpretability in Parameter Space: Minimizing Mechanistic Description Length with Attribution-based Parameter Decomposition","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Dan Braun, Jake Mendel, Lee Sharkey, Lucius Bushnaq, Stefan Heimersheim","submitted_at":"2025-01-24T21:31:12Z","abstract_excerpt":"Mechanistic interpretability aims to understand the internal mechanisms learned by neural networks. Despite recent progress toward this goal, it remains unclear how best to decompose neural network parameters into mechanistic components. We introduce Attribution-based Parameter Decomposition (APD), a method that directly decomposes a neural network's parameters into components that (i) are faithful to the parameters of the original network, (ii) require a minimal number of components to process any input, and (iii) are maximally simple. Our approach thus optimizes for a minimal length descript"},"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":"2501.14926","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-01-24T21:31:12Z","cross_cats_sorted":["stat.ML"],"title_canon_sha256":"e6b8991209d0259076347c28e03afa1ac2f122a838660fa644b5da50d14b7f38","abstract_canon_sha256":"0949fcc6cfa4eb826cb3b1d23eb3a4377de3e1d8cbd9bbbb926a86badfa43d4c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:11:32.110926Z","signature_b64":"sqhjrgYJJTALGScZunk6ap3TBz0xhcswWOsOx2TBqdiIFwiDlET5e+FnyXBP+ZWgf8ymsk09ZTWte9KCMbhcCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3e63e002a99684fa5396d302e73e960a6983f48a23d97f7d2e2208f5a0bad5aa","last_reissued_at":"2026-07-05T10:11:32.110488Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:11:32.110488Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Interpretability in Parameter Space: Minimizing Mechanistic Description Length with Attribution-based Parameter Decomposition","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["stat.ML"],"primary_cat":"cs.LG","authors_text":"Dan Braun, Jake Mendel, Lee Sharkey, Lucius Bushnaq, Stefan Heimersheim","submitted_at":"2025-01-24T21:31:12Z","abstract_excerpt":"Mechanistic interpretability aims to understand the internal mechanisms learned by neural networks. Despite recent progress toward this goal, it remains unclear how best to decompose neural network parameters into mechanistic components. We introduce Attribution-based Parameter Decomposition (APD), a method that directly decomposes a neural network's parameters into components that (i) are faithful to the parameters of the original network, (ii) require a minimal number of components to process any input, and (iii) are maximally simple. Our approach thus optimizes for a minimal length descript"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.14926","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/2501.14926/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":"2501.14926","created_at":"2026-07-05T10:11:32.110548+00:00"},{"alias_kind":"arxiv_version","alias_value":"2501.14926v4","created_at":"2026-07-05T10:11:32.110548+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.14926","created_at":"2026-07-05T10:11:32.110548+00:00"},{"alias_kind":"pith_short_12","alias_value":"HZR6AAVJS2CP","created_at":"2026-07-05T10:11:32.110548+00:00"},{"alias_kind":"pith_short_16","alias_value":"HZR6AAVJS2CPUU4W","created_at":"2026-07-05T10:11:32.110548+00:00"},{"alias_kind":"pith_short_8","alias_value":"HZR6AAVJ","created_at":"2026-07-05T10:11:32.110548+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":7,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.08934","citing_title":"From Mechanistic to Compositional Interpretability","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12770","citing_title":"WriteSAE: Sparse Autoencoders for Recurrent State","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12770","citing_title":"WriteSAE: Sparse Autoencoders for Recurrent State","ref_index":96,"is_internal_anchor":false},{"citing_arxiv_id":"2509.06701","citing_title":"Probabilistic Modeling of Latent Agentic Substructures in Deep Neural Networks","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12770","citing_title":"WriteSAE: Sparse Autoencoders for Recurrent State","ref_index":96,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12770","citing_title":"WriteSAE: Sparse Autoencoders for Recurrent State","ref_index":96,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08934","citing_title":"From Mechanistic to Compositional Interpretability","ref_index":165,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/HZR6AAVJS2CPUU4W2MBOOPUWBJ","json":"https://pith.science/pith/HZR6AAVJS2CPUU4W2MBOOPUWBJ.json","graph_json":"https://pith.science/api/pith-number/HZR6AAVJS2CPUU4W2MBOOPUWBJ/graph.json","events_json":"https://pith.science/api/pith-number/HZR6AAVJS2CPUU4W2MBOOPUWBJ/events.json","paper":"https://pith.science/paper/HZR6AAVJ"},"agent_actions":{"view_html":"https://pith.science/pith/HZR6AAVJS2CPUU4W2MBOOPUWBJ","download_json":"https://pith.science/pith/HZR6AAVJS2CPUU4W2MBOOPUWBJ.json","view_paper":"https://pith.science/paper/HZR6AAVJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2501.14926&json=true","fetch_graph":"https://pith.science/api/pith-number/HZR6AAVJS2CPUU4W2MBOOPUWBJ/graph.json","fetch_events":"https://pith.science/api/pith-number/HZR6AAVJS2CPUU4W2MBOOPUWBJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HZR6AAVJS2CPUU4W2MBOOPUWBJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HZR6AAVJS2CPUU4W2MBOOPUWBJ/action/storage_attestation","attest_author":"https://pith.science/pith/HZR6AAVJS2CPUU4W2MBOOPUWBJ/action/author_attestation","sign_citation":"https://pith.science/pith/HZR6AAVJS2CPUU4W2MBOOPUWBJ/action/citation_signature","submit_replication":"https://pith.science/pith/HZR6AAVJS2CPUU4W2MBOOPUWBJ/action/replication_record"}},"created_at":"2026-07-05T10:11:32.110548+00:00","updated_at":"2026-07-05T10:11:32.110548+00:00"}