{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:A6BJG3MDFGRHVEARUPMICE2PCY","short_pith_number":"pith:A6BJG3MD","schema_version":"1.0","canonical_sha256":"0782936d8329a27a9011a3d881134f161cb9d7d73301a570cc0c14a08be6a1e4","source":{"kind":"arxiv","id":"2606.08496","version":1},"attestation_state":"computed","paper":{"title":"SAEExplainer: Interpreting SAE Features with Activation-Guided Preference Optimization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Fei Sun, Haiyan Zhao, Jingyi He, Mengnan Du, Ruxue Shi, Xin Wang, Yanguang Liu","submitted_at":"2026-06-07T07:54:25Z","abstract_excerpt":"Although Sparse Autoencoders (SAEs) have mitigated the opacity of large language models (LLMs) by decomposing dense representations into sparse features, explaining these features still remains a central challenge. Current explanation methods, however, typically operate within an open-loop paradigm, failing to leverage mechanistic feedback for further refinement. In this paper, we propose SAEExplainer, a training framework utilizes activation scores as an objective reward signal to train the model for self-correction and iterative bootstrapping. By iteratively verifying and correcting foundati"},"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":"2606.08496","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-06-07T07:54:25Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"47d77fbb04572837095ff536c087d4867f891f2d7dbe86dbf7c143911fa0037e","abstract_canon_sha256":"972a57118a72b69185af5e21234e99ddefc0cb67cfb9160c736bc23c372c1bb7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-09T01:05:38.339971Z","signature_b64":"lWs5k3rJFLyKw60PT5pZuh1OK9rOiqouPUBB2TtW/0vfCQplGHaxbvwV6MV+5GPiPs8cAXNBr7w208yST5JeBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0782936d8329a27a9011a3d881134f161cb9d7d73301a570cc0c14a08be6a1e4","last_reissued_at":"2026-06-09T01:05:38.339533Z","signature_status":"signed_v1","first_computed_at":"2026-06-09T01:05:38.339533Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SAEExplainer: Interpreting SAE Features with Activation-Guided Preference Optimization","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Fei Sun, Haiyan Zhao, Jingyi He, Mengnan Du, Ruxue Shi, Xin Wang, Yanguang Liu","submitted_at":"2026-06-07T07:54:25Z","abstract_excerpt":"Although Sparse Autoencoders (SAEs) have mitigated the opacity of large language models (LLMs) by decomposing dense representations into sparse features, explaining these features still remains a central challenge. Current explanation methods, however, typically operate within an open-loop paradigm, failing to leverage mechanistic feedback for further refinement. In this paper, we propose SAEExplainer, a training framework utilizes activation scores as an objective reward signal to train the model for self-correction and iterative bootstrapping. By iteratively verifying and correcting foundati"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.08496","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/2606.08496/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":"2606.08496","created_at":"2026-06-09T01:05:38.339603+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.08496v1","created_at":"2026-06-09T01:05:38.339603+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.08496","created_at":"2026-06-09T01:05:38.339603+00:00"},{"alias_kind":"pith_short_12","alias_value":"A6BJG3MDFGRH","created_at":"2026-06-09T01:05:38.339603+00:00"},{"alias_kind":"pith_short_16","alias_value":"A6BJG3MDFGRHVEAR","created_at":"2026-06-09T01:05:38.339603+00:00"},{"alias_kind":"pith_short_8","alias_value":"A6BJG3MD","created_at":"2026-06-09T01:05:38.339603+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/A6BJG3MDFGRHVEARUPMICE2PCY","json":"https://pith.science/pith/A6BJG3MDFGRHVEARUPMICE2PCY.json","graph_json":"https://pith.science/api/pith-number/A6BJG3MDFGRHVEARUPMICE2PCY/graph.json","events_json":"https://pith.science/api/pith-number/A6BJG3MDFGRHVEARUPMICE2PCY/events.json","paper":"https://pith.science/paper/A6BJG3MD"},"agent_actions":{"view_html":"https://pith.science/pith/A6BJG3MDFGRHVEARUPMICE2PCY","download_json":"https://pith.science/pith/A6BJG3MDFGRHVEARUPMICE2PCY.json","view_paper":"https://pith.science/paper/A6BJG3MD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.08496&json=true","fetch_graph":"https://pith.science/api/pith-number/A6BJG3MDFGRHVEARUPMICE2PCY/graph.json","fetch_events":"https://pith.science/api/pith-number/A6BJG3MDFGRHVEARUPMICE2PCY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/A6BJG3MDFGRHVEARUPMICE2PCY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/A6BJG3MDFGRHVEARUPMICE2PCY/action/storage_attestation","attest_author":"https://pith.science/pith/A6BJG3MDFGRHVEARUPMICE2PCY/action/author_attestation","sign_citation":"https://pith.science/pith/A6BJG3MDFGRHVEARUPMICE2PCY/action/citation_signature","submit_replication":"https://pith.science/pith/A6BJG3MDFGRHVEARUPMICE2PCY/action/replication_record"}},"created_at":"2026-06-09T01:05:38.339603+00:00","updated_at":"2026-06-09T01:05:38.339603+00:00"}