{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:V52XFN7KXLSXVTMAEUQGKN7S77","short_pith_number":"pith:V52XFN7K","schema_version":"1.0","canonical_sha256":"af7572b7eabae57acd8025206537f2ffe2ca1b16109a3635df98b0cee5be6db4","source":{"kind":"arxiv","id":"2406.04028","version":1},"attestation_state":"computed","paper":{"title":"Contrastive Sparse Autoencoders for Interpreting Planning of Chess-Playing Agents","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Yoann Poupart","submitted_at":"2024-06-06T12:57:31Z","abstract_excerpt":"AI led chess systems to a superhuman level, yet these systems heavily rely on black-box algorithms. This is unsustainable in ensuring transparency to the end-user, particularly when these systems are responsible for sensitive decision-making. Recent interpretability work has shown that the inner representations of Deep Neural Networks (DNNs) were fathomable and contained human-understandable concepts. Yet, these methods are seldom contextualised and are often based on a single hidden state, which makes them unable to interpret multi-step reasoning, e.g. planning. In this respect, we propose co"},"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":"2406.04028","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2024-06-06T12:57:31Z","cross_cats_sorted":[],"title_canon_sha256":"6d3425267df355c307e47ea26ffde10b920d268c45790c56d747b18d6d626e7a","abstract_canon_sha256":"7bf5068311c541cf7e53a061e47aee44a109a8c3b341814920d26213d6b06a19"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:30:13.650475Z","signature_b64":"9X2txmCjrFIjVuJ49YrdWDoPb7haZS27jsUEJ3oo9Trr4GKy69wnVZVBE1D/Y94PjVzxmAW0pRfyKPfHAm4FBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"af7572b7eabae57acd8025206537f2ffe2ca1b16109a3635df98b0cee5be6db4","last_reissued_at":"2026-07-05T08:30:13.649984Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:30:13.649984Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Contrastive Sparse Autoencoders for Interpreting Planning of Chess-Playing Agents","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Yoann Poupart","submitted_at":"2024-06-06T12:57:31Z","abstract_excerpt":"AI led chess systems to a superhuman level, yet these systems heavily rely on black-box algorithms. This is unsustainable in ensuring transparency to the end-user, particularly when these systems are responsible for sensitive decision-making. Recent interpretability work has shown that the inner representations of Deep Neural Networks (DNNs) were fathomable and contained human-understandable concepts. Yet, these methods are seldom contextualised and are often based on a single hidden state, which makes them unable to interpret multi-step reasoning, e.g. planning. In this respect, we propose co"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.04028","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/2406.04028/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":"2406.04028","created_at":"2026-07-05T08:30:13.650038+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.04028v1","created_at":"2026-07-05T08:30:13.650038+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.04028","created_at":"2026-07-05T08:30:13.650038+00:00"},{"alias_kind":"pith_short_12","alias_value":"V52XFN7KXLSX","created_at":"2026-07-05T08:30:13.650038+00:00"},{"alias_kind":"pith_short_16","alias_value":"V52XFN7KXLSXVTMA","created_at":"2026-07-05T08:30:13.650038+00:00"},{"alias_kind":"pith_short_8","alias_value":"V52XFN7K","created_at":"2026-07-05T08:30:13.650038+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2502.00726","citing_title":"Perspectives for Direct Interpretability in Multi-Agent Deep Reinforcement Learning","ref_index":92,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/V52XFN7KXLSXVTMAEUQGKN7S77","json":"https://pith.science/pith/V52XFN7KXLSXVTMAEUQGKN7S77.json","graph_json":"https://pith.science/api/pith-number/V52XFN7KXLSXVTMAEUQGKN7S77/graph.json","events_json":"https://pith.science/api/pith-number/V52XFN7KXLSXVTMAEUQGKN7S77/events.json","paper":"https://pith.science/paper/V52XFN7K"},"agent_actions":{"view_html":"https://pith.science/pith/V52XFN7KXLSXVTMAEUQGKN7S77","download_json":"https://pith.science/pith/V52XFN7KXLSXVTMAEUQGKN7S77.json","view_paper":"https://pith.science/paper/V52XFN7K","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.04028&json=true","fetch_graph":"https://pith.science/api/pith-number/V52XFN7KXLSXVTMAEUQGKN7S77/graph.json","fetch_events":"https://pith.science/api/pith-number/V52XFN7KXLSXVTMAEUQGKN7S77/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/V52XFN7KXLSXVTMAEUQGKN7S77/action/timestamp_anchor","attest_storage":"https://pith.science/pith/V52XFN7KXLSXVTMAEUQGKN7S77/action/storage_attestation","attest_author":"https://pith.science/pith/V52XFN7KXLSXVTMAEUQGKN7S77/action/author_attestation","sign_citation":"https://pith.science/pith/V52XFN7KXLSXVTMAEUQGKN7S77/action/citation_signature","submit_replication":"https://pith.science/pith/V52XFN7KXLSXVTMAEUQGKN7S77/action/replication_record"}},"created_at":"2026-07-05T08:30:13.650038+00:00","updated_at":"2026-07-05T08:30:13.650038+00:00"}