{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:UMF2VQRA2BSDMGXQU6EXQGBUZT","short_pith_number":"pith:UMF2VQRA","canonical_record":{"source":{"id":"2501.18052","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-01-29T23:29:47Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"0b690fedcc92daf3c9c9795e2fdb3cffcdfae59eab91a0a1753e394cec90ed91","abstract_canon_sha256":"a0c5b4e4bd30bd9c44953d14302e3c2ba5b3fc779de3f889812a790acb728d0b"},"schema_version":"1.0"},"canonical_sha256":"a30baac220d064361af0a789781834ccd52e3b9dfdb10c73ac8a25f37f88abd5","source":{"kind":"arxiv","id":"2501.18052","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2501.18052","created_at":"2026-07-05T11:07:19Z"},{"alias_kind":"arxiv_version","alias_value":"2501.18052v3","created_at":"2026-07-05T11:07:19Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.18052","created_at":"2026-07-05T11:07:19Z"},{"alias_kind":"pith_short_12","alias_value":"UMF2VQRA2BSD","created_at":"2026-07-05T11:07:19Z"},{"alias_kind":"pith_short_16","alias_value":"UMF2VQRA2BSDMGXQ","created_at":"2026-07-05T11:07:19Z"},{"alias_kind":"pith_short_8","alias_value":"UMF2VQRA","created_at":"2026-07-05T11:07:19Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:UMF2VQRA2BSDMGXQU6EXQGBUZT","target":"record","payload":{"canonical_record":{"source":{"id":"2501.18052","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-01-29T23:29:47Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"0b690fedcc92daf3c9c9795e2fdb3cffcdfae59eab91a0a1753e394cec90ed91","abstract_canon_sha256":"a0c5b4e4bd30bd9c44953d14302e3c2ba5b3fc779de3f889812a790acb728d0b"},"schema_version":"1.0"},"canonical_sha256":"a30baac220d064361af0a789781834ccd52e3b9dfdb10c73ac8a25f37f88abd5","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:07:19.743871Z","signature_b64":"ogf/nXUaG1UPb09ikBJl+5GTINvtl/zx7IkcIQfo6egaNu1+DDY412IeiKxJ7XvMd1VGA51MYJcpQTnLwJi4Cg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a30baac220d064361af0a789781834ccd52e3b9dfdb10c73ac8a25f37f88abd5","last_reissued_at":"2026-07-05T11:07:19.743395Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:07:19.743395Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2501.18052","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T11:07:19Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"1Fqns3PwLE2VLKSqxGjXUQrCsH6uweMXifMp/bwi2rzvGxJzWNjccODP0irhOBr+PODJ+T44XJ7cKyPJVqANBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-23T01:17:25.747141Z"},"content_sha256":"8d0143b7b1cb72b2da0a103411280c0ff2e1bb3bc30cc12f93a70b5a00da00c1","schema_version":"1.0","event_id":"sha256:8d0143b7b1cb72b2da0a103411280c0ff2e1bb3bc30cc12f93a70b5a00da00c1"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:UMF2VQRA2BSDMGXQU6EXQGBUZT","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"SAeUron: Interpretable Concept Unlearning in Diffusion Models with Sparse Autoencoders","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Bartosz Cywi\\'nski, Kamil Deja","submitted_at":"2025-01-29T23:29:47Z","abstract_excerpt":"Diffusion models, while powerful, can inadvertently generate harmful or undesirable content, raising significant ethical and safety concerns. Recent machine unlearning approaches offer potential solutions but often lack transparency, making it difficult to understand the changes they introduce to the base model. In this work, we introduce SAeUron, a novel method leveraging features learned by sparse autoencoders (SAEs) to remove unwanted concepts in text-to-image diffusion models. First, we demonstrate that SAEs, trained in an unsupervised manner on activations from multiple denoising timestep"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.18052","kind":"arxiv","version":3},"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/2501.18052/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":"2026-07-05T11:07:19Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"4dVtG3myQXD9i7R2X7NE6/YZGzbpBrutjN2HXMaTo9dyR3I8BoCCiMGvxnI+upQJawbxzCeiePMNCdn+kDwhAg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-23T01:17:25.748009Z"},"content_sha256":"84314a080b354d6adfd422e1a9709131b6036336a703bf90226642f79139b192","schema_version":"1.0","event_id":"sha256:84314a080b354d6adfd422e1a9709131b6036336a703bf90226642f79139b192"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/UMF2VQRA2BSDMGXQU6EXQGBUZT/bundle.json","state_url":"https://pith.science/pith/UMF2VQRA2BSDMGXQU6EXQGBUZT/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/UMF2VQRA2BSDMGXQU6EXQGBUZT/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":"2026-08-23T01:17:25Z","links":{"resolver":"https://pith.science/pith/UMF2VQRA2BSDMGXQU6EXQGBUZT","bundle":"https://pith.science/pith/UMF2VQRA2BSDMGXQU6EXQGBUZT/bundle.json","state":"https://pith.science/pith/UMF2VQRA2BSDMGXQU6EXQGBUZT/state.json","well_known_bundle":"https://pith.science/.well-known/pith/UMF2VQRA2BSDMGXQU6EXQGBUZT/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:UMF2VQRA2BSDMGXQU6EXQGBUZT","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":"a0c5b4e4bd30bd9c44953d14302e3c2ba5b3fc779de3f889812a790acb728d0b","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-01-29T23:29:47Z","title_canon_sha256":"0b690fedcc92daf3c9c9795e2fdb3cffcdfae59eab91a0a1753e394cec90ed91"},"schema_version":"1.0","source":{"id":"2501.18052","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2501.18052","created_at":"2026-07-05T11:07:19Z"},{"alias_kind":"arxiv_version","alias_value":"2501.18052v3","created_at":"2026-07-05T11:07:19Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.18052","created_at":"2026-07-05T11:07:19Z"},{"alias_kind":"pith_short_12","alias_value":"UMF2VQRA2BSD","created_at":"2026-07-05T11:07:19Z"},{"alias_kind":"pith_short_16","alias_value":"UMF2VQRA2BSDMGXQ","created_at":"2026-07-05T11:07:19Z"},{"alias_kind":"pith_short_8","alias_value":"UMF2VQRA","created_at":"2026-07-05T11:07:19Z"}],"graph_snapshots":[{"event_id":"sha256:84314a080b354d6adfd422e1a9709131b6036336a703bf90226642f79139b192","target":"graph","created_at":"2026-07-05T11:07:19Z","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/2501.18052/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Diffusion models, while powerful, can inadvertently generate harmful or undesirable content, raising significant ethical and safety concerns. Recent machine unlearning approaches offer potential solutions but often lack transparency, making it difficult to understand the changes they introduce to the base model. In this work, we introduce SAeUron, a novel method leveraging features learned by sparse autoencoders (SAEs) to remove unwanted concepts in text-to-image diffusion models. First, we demonstrate that SAEs, trained in an unsupervised manner on activations from multiple denoising timestep","authors_text":"Bartosz Cywi\\'nski, Kamil Deja","cross_cats":["cs.AI"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-01-29T23:29:47Z","title":"SAeUron: Interpretable Concept Unlearning in Diffusion Models with Sparse Autoencoders"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.18052","kind":"arxiv","version":3},"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":"sha256:8d0143b7b1cb72b2da0a103411280c0ff2e1bb3bc30cc12f93a70b5a00da00c1","target":"record","created_at":"2026-07-05T11:07:19Z","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":"a0c5b4e4bd30bd9c44953d14302e3c2ba5b3fc779de3f889812a790acb728d0b","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-01-29T23:29:47Z","title_canon_sha256":"0b690fedcc92daf3c9c9795e2fdb3cffcdfae59eab91a0a1753e394cec90ed91"},"schema_version":"1.0","source":{"id":"2501.18052","kind":"arxiv","version":3}},"canonical_sha256":"a30baac220d064361af0a789781834ccd52e3b9dfdb10c73ac8a25f37f88abd5","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"a30baac220d064361af0a789781834ccd52e3b9dfdb10c73ac8a25f37f88abd5","first_computed_at":"2026-07-05T11:07:19.743395Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T11:07:19.743395Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"ogf/nXUaG1UPb09ikBJl+5GTINvtl/zx7IkcIQfo6egaNu1+DDY412IeiKxJ7XvMd1VGA51MYJcpQTnLwJi4Cg==","signature_status":"signed_v1","signed_at":"2026-07-05T11:07:19.743871Z","signed_message":"canonical_sha256_bytes"},"source_id":"2501.18052","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:8d0143b7b1cb72b2da0a103411280c0ff2e1bb3bc30cc12f93a70b5a00da00c1","sha256:84314a080b354d6adfd422e1a9709131b6036336a703bf90226642f79139b192"],"state_sha256":"afc1ae4d42e70c7f5e230299c1561628ccdcf11606ae3c8929e40fe369792081"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"W02lQkpQnFgVdnhhh/WHxoCgT82yricFYIdaIGq5Eg6Uvsl/jXXBVwkW8MPiEiDaz/RleX8UB/w6buFwNZzRBQ==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-23T01:17:25.753274Z","bundle_sha256":"46d1f13836f7ad452fd812b66bb2ea6052d0060eb08c3709184fe76a88c94470"}}