{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:A6BJG3MDFGRHVEARUPMICE2PCY","short_pith_number":"pith:A6BJG3MD","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"},"canonical_sha256":"0782936d8329a27a9011a3d881134f161cb9d7d73301a570cc0c14a08be6a1e4","source":{"kind":"arxiv","id":"2606.08496","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.08496","created_at":"2026-06-09T01:05:38Z"},{"alias_kind":"arxiv_version","alias_value":"2606.08496v1","created_at":"2026-06-09T01:05:38Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.08496","created_at":"2026-06-09T01:05:38Z"},{"alias_kind":"pith_short_12","alias_value":"A6BJG3MDFGRH","created_at":"2026-06-09T01:05:38Z"},{"alias_kind":"pith_short_16","alias_value":"A6BJG3MDFGRHVEAR","created_at":"2026-06-09T01:05:38Z"},{"alias_kind":"pith_short_8","alias_value":"A6BJG3MD","created_at":"2026-06-09T01:05:38Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:A6BJG3MDFGRHVEARUPMICE2PCY","target":"record","payload":{"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"},"canonical_sha256":"0782936d8329a27a9011a3d881134f161cb9d7d73301a570cc0c14a08be6a1e4","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"},"source_kind":"arxiv","source_id":"2606.08496","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-06-09T01:05:38Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Zd0ShGewsn2h/Cj4gSSj8llMOsYieM1gPKSbBcPL14/BjKOyokn6amIRoh0SsfP4vvDieOKlk9JsTjMVle6jDw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-02T12:15:25.653437Z"},"content_sha256":"5793fd2ed22ff46f251587b06abf468dc174beae0b68ab4a506c25bb286724e6","schema_version":"1.0","event_id":"sha256:5793fd2ed22ff46f251587b06abf468dc174beae0b68ab4a506c25bb286724e6"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:A6BJG3MDFGRHVEARUPMICE2PCY","target":"graph","payload":{"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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-06-09T01:05:38Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"uupIQB69UPiMG2ns65317KG1UelxC+JwNPL6yUgTs3V3kVCWFyiB1nHwxgQtT9AVEH8JgquQ1bI5aW78d23YBQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-02T12:15:25.653806Z"},"content_sha256":"bf1f4cba2d438acf1f853ac5c0a8725c2196c59f9bb38a76e9ea01d219bf9d9c","schema_version":"1.0","event_id":"sha256:bf1f4cba2d438acf1f853ac5c0a8725c2196c59f9bb38a76e9ea01d219bf9d9c"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/A6BJG3MDFGRHVEARUPMICE2PCY/bundle.json","state_url":"https://pith.science/pith/A6BJG3MDFGRHVEARUPMICE2PCY/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/A6BJG3MDFGRHVEARUPMICE2PCY/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-07-02T12:15:25Z","links":{"resolver":"https://pith.science/pith/A6BJG3MDFGRHVEARUPMICE2PCY","bundle":"https://pith.science/pith/A6BJG3MDFGRHVEARUPMICE2PCY/bundle.json","state":"https://pith.science/pith/A6BJG3MDFGRHVEARUPMICE2PCY/state.json","well_known_bundle":"https://pith.science/.well-known/pith/A6BJG3MDFGRHVEARUPMICE2PCY/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:A6BJG3MDFGRHVEARUPMICE2PCY","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"972a57118a72b69185af5e21234e99ddefc0cb67cfb9160c736bc23c372c1bb7","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-06-07T07:54:25Z","title_canon_sha256":"47d77fbb04572837095ff536c087d4867f891f2d7dbe86dbf7c143911fa0037e"},"schema_version":"1.0","source":{"id":"2606.08496","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2606.08496","created_at":"2026-06-09T01:05:38Z"},{"alias_kind":"arxiv_version","alias_value":"2606.08496v1","created_at":"2026-06-09T01:05:38Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.08496","created_at":"2026-06-09T01:05:38Z"},{"alias_kind":"pith_short_12","alias_value":"A6BJG3MDFGRH","created_at":"2026-06-09T01:05:38Z"},{"alias_kind":"pith_short_16","alias_value":"A6BJG3MDFGRHVEAR","created_at":"2026-06-09T01:05:38Z"},{"alias_kind":"pith_short_8","alias_value":"A6BJG3MD","created_at":"2026-06-09T01:05:38Z"}],"graph_snapshots":[{"event_id":"sha256:bf1f4cba2d438acf1f853ac5c0a8725c2196c59f9bb38a76e9ea01d219bf9d9c","target":"graph","created_at":"2026-06-09T01:05:38Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2606.08496/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"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","authors_text":"Fei Sun, Haiyan Zhao, Jingyi He, Mengnan Du, Ruxue Shi, Xin Wang, Yanguang Liu","cross_cats":["cs.LG"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-06-07T07:54:25Z","title":"SAEExplainer: Interpreting SAE Features with Activation-Guided Preference Optimization"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.08496","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:5793fd2ed22ff46f251587b06abf468dc174beae0b68ab4a506c25bb286724e6","target":"record","created_at":"2026-06-09T01:05:38Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"972a57118a72b69185af5e21234e99ddefc0cb67cfb9160c736bc23c372c1bb7","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-06-07T07:54:25Z","title_canon_sha256":"47d77fbb04572837095ff536c087d4867f891f2d7dbe86dbf7c143911fa0037e"},"schema_version":"1.0","source":{"id":"2606.08496","kind":"arxiv","version":1}},"canonical_sha256":"0782936d8329a27a9011a3d881134f161cb9d7d73301a570cc0c14a08be6a1e4","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"0782936d8329a27a9011a3d881134f161cb9d7d73301a570cc0c14a08be6a1e4","first_computed_at":"2026-06-09T01:05:38.339533Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-09T01:05:38.339533Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"lWs5k3rJFLyKw60PT5pZuh1OK9rOiqouPUBB2TtW/0vfCQplGHaxbvwV6MV+5GPiPs8cAXNBr7w208yST5JeBA==","signature_status":"signed_v1","signed_at":"2026-06-09T01:05:38.339971Z","signed_message":"canonical_sha256_bytes"},"source_id":"2606.08496","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:5793fd2ed22ff46f251587b06abf468dc174beae0b68ab4a506c25bb286724e6","sha256:bf1f4cba2d438acf1f853ac5c0a8725c2196c59f9bb38a76e9ea01d219bf9d9c"],"state_sha256":"f37dff478c16ffc28ccf1aa1026f10417f520e887d521f6437bdc6add715e1f8"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Zje28DaDL6dPc2SldSGlR35sD2Nr09pxUi+c7u5j6zEzZtSSaNuXz+7i49s1TlVd252pBddlQTl1yhOdQ/OxDg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-02T12:15:25.655873Z","bundle_sha256":"2f5f66a6fd8447f0bfd70f5cd1a52e29d52dd53fff575035b6effcba15740f77"}}