{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2022:JOO47KF2SMPHAQGDUXVQI75L2N","short_pith_number":"pith:JOO47KF2","canonical_record":{"source":{"id":"2211.09169","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2022-11-16T19:32:43Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"bfd91cbb0da5c34427f4bfa9858718fdebba6c82bb6c39b611fba4763248748f","abstract_canon_sha256":"58fc0602bb2256abc076b10c597f94b7a74428a5c587ee41511ed9443c1df50d"},"schema_version":"1.0"},"canonical_sha256":"4b9dcfa8ba931e7040c3a5eb047fabd341c50907444055986dc9ce6c90e33eda","source":{"kind":"arxiv","id":"2211.09169","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2211.09169","created_at":"2026-07-05T05:16:53Z"},{"alias_kind":"arxiv_version","alias_value":"2211.09169v1","created_at":"2026-07-05T05:16:53Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2211.09169","created_at":"2026-07-05T05:16:53Z"},{"alias_kind":"pith_short_12","alias_value":"JOO47KF2SMPH","created_at":"2026-07-05T05:16:53Z"},{"alias_kind":"pith_short_16","alias_value":"JOO47KF2SMPHAQGD","created_at":"2026-07-05T05:16:53Z"},{"alias_kind":"pith_short_8","alias_value":"JOO47KF2","created_at":"2026-07-05T05:16:53Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2022:JOO47KF2SMPHAQGDUXVQI75L2N","target":"record","payload":{"canonical_record":{"source":{"id":"2211.09169","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2022-11-16T19:32:43Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"bfd91cbb0da5c34427f4bfa9858718fdebba6c82bb6c39b611fba4763248748f","abstract_canon_sha256":"58fc0602bb2256abc076b10c597f94b7a74428a5c587ee41511ed9443c1df50d"},"schema_version":"1.0"},"canonical_sha256":"4b9dcfa8ba931e7040c3a5eb047fabd341c50907444055986dc9ce6c90e33eda","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:16:53.409805Z","signature_b64":"qG+5loiq2MM8oyD3qc8jDXBA4ZHIDGUTboc9UcInfIXn3txE7R79vlCfhQiVZVsIirqCoxFWTdV0qYjRfNCPAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4b9dcfa8ba931e7040c3a5eb047fabd341c50907444055986dc9ce6c90e33eda","last_reissued_at":"2026-07-05T05:16:53.409409Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:16:53.409409Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2211.09169","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":"2026-07-05T05:16:53Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"dQ6cE4+6tsjL1Fa/EX2SDh49gVw6wVG14pc0Jc2UBNWCDXxd5+0G9Ki7BfZbdEqX7q+Bm/kGF4sYAJYYwN4kAg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-05T23:13:17.765061Z"},"content_sha256":"df4d35cab3fc8b7a2023e6cf2caa0b8337ea1eeea298c7bc93e3ca8c7d464a4a","schema_version":"1.0","event_id":"sha256:df4d35cab3fc8b7a2023e6cf2caa0b8337ea1eeea298c7bc93e3ca8c7d464a4a"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2022:JOO47KF2SMPHAQGDUXVQI75L2N","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Engineering Monosemanticity in Toy Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Adam S. Jermyn, Evan Hubinger, Nicholas Schiefer","submitted_at":"2022-11-16T19:32:43Z","abstract_excerpt":"In some neural networks, individual neurons correspond to natural ``features'' in the input. Such \\emph{monosemantic} neurons are of great help in interpretability studies, as they can be cleanly understood. In this work we report preliminary attempts to engineer monosemanticity in toy models. We find that models can be made more monosemantic without increasing the loss by just changing which local minimum the training process finds. More monosemantic loss minima have moderate negative biases, and we are able to use this fact to engineer highly monosemantic models. We are able to mechanistical"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2211.09169","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/2211.09169/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":"2026-07-05T05:16:53Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"3bfFO8bv4002ovXKwRGwEsmW81vCNjpEqY0+kvkbgI8r9g+Sgh4agPWaOc9RNRa2DGMRSqouwnKIaBx45MJgDw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-05T23:13:17.766018Z"},"content_sha256":"7447a00cdcf441149afdc434c41a23ffbcf93278d634ecd9fd1cef5d24e7ea48","schema_version":"1.0","event_id":"sha256:7447a00cdcf441149afdc434c41a23ffbcf93278d634ecd9fd1cef5d24e7ea48"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/JOO47KF2SMPHAQGDUXVQI75L2N/bundle.json","state_url":"https://pith.science/pith/JOO47KF2SMPHAQGDUXVQI75L2N/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/JOO47KF2SMPHAQGDUXVQI75L2N/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":"2026-08-05T23:13:17Z","links":{"resolver":"https://pith.science/pith/JOO47KF2SMPHAQGDUXVQI75L2N","bundle":"https://pith.science/pith/JOO47KF2SMPHAQGDUXVQI75L2N/bundle.json","state":"https://pith.science/pith/JOO47KF2SMPHAQGDUXVQI75L2N/state.json","well_known_bundle":"https://pith.science/.well-known/pith/JOO47KF2SMPHAQGDUXVQI75L2N/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2022:JOO47KF2SMPHAQGDUXVQI75L2N","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":"58fc0602bb2256abc076b10c597f94b7a74428a5c587ee41511ed9443c1df50d","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2022-11-16T19:32:43Z","title_canon_sha256":"bfd91cbb0da5c34427f4bfa9858718fdebba6c82bb6c39b611fba4763248748f"},"schema_version":"1.0","source":{"id":"2211.09169","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2211.09169","created_at":"2026-07-05T05:16:53Z"},{"alias_kind":"arxiv_version","alias_value":"2211.09169v1","created_at":"2026-07-05T05:16:53Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2211.09169","created_at":"2026-07-05T05:16:53Z"},{"alias_kind":"pith_short_12","alias_value":"JOO47KF2SMPH","created_at":"2026-07-05T05:16:53Z"},{"alias_kind":"pith_short_16","alias_value":"JOO47KF2SMPHAQGD","created_at":"2026-07-05T05:16:53Z"},{"alias_kind":"pith_short_8","alias_value":"JOO47KF2","created_at":"2026-07-05T05:16:53Z"}],"graph_snapshots":[{"event_id":"sha256:7447a00cdcf441149afdc434c41a23ffbcf93278d634ecd9fd1cef5d24e7ea48","target":"graph","created_at":"2026-07-05T05:16:53Z","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/2211.09169/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"In some neural networks, individual neurons correspond to natural ``features'' in the input. Such \\emph{monosemantic} neurons are of great help in interpretability studies, as they can be cleanly understood. In this work we report preliminary attempts to engineer monosemanticity in toy models. We find that models can be made more monosemantic without increasing the loss by just changing which local minimum the training process finds. More monosemantic loss minima have moderate negative biases, and we are able to use this fact to engineer highly monosemantic models. We are able to mechanistical","authors_text":"Adam S. Jermyn, Evan Hubinger, Nicholas Schiefer","cross_cats":["cs.AI"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2022-11-16T19:32:43Z","title":"Engineering Monosemanticity in Toy Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2211.09169","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:df4d35cab3fc8b7a2023e6cf2caa0b8337ea1eeea298c7bc93e3ca8c7d464a4a","target":"record","created_at":"2026-07-05T05:16:53Z","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":"58fc0602bb2256abc076b10c597f94b7a74428a5c587ee41511ed9443c1df50d","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2022-11-16T19:32:43Z","title_canon_sha256":"bfd91cbb0da5c34427f4bfa9858718fdebba6c82bb6c39b611fba4763248748f"},"schema_version":"1.0","source":{"id":"2211.09169","kind":"arxiv","version":1}},"canonical_sha256":"4b9dcfa8ba931e7040c3a5eb047fabd341c50907444055986dc9ce6c90e33eda","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"4b9dcfa8ba931e7040c3a5eb047fabd341c50907444055986dc9ce6c90e33eda","first_computed_at":"2026-07-05T05:16:53.409409Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T05:16:53.409409Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"qG+5loiq2MM8oyD3qc8jDXBA4ZHIDGUTboc9UcInfIXn3txE7R79vlCfhQiVZVsIirqCoxFWTdV0qYjRfNCPAw==","signature_status":"signed_v1","signed_at":"2026-07-05T05:16:53.409805Z","signed_message":"canonical_sha256_bytes"},"source_id":"2211.09169","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:df4d35cab3fc8b7a2023e6cf2caa0b8337ea1eeea298c7bc93e3ca8c7d464a4a","sha256:7447a00cdcf441149afdc434c41a23ffbcf93278d634ecd9fd1cef5d24e7ea48"],"state_sha256":"4ca96a0ef16cbe7c4f7a355591dc8f5bcb995fabbf3a15f6036c6e06f02d5555"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"6iPv6g13jkvQVeiEbftK8+1g9d9yscd5daZq6Wi5o8gnKIh4F8/S0kTK/x/gvoUiykwKcfxZKsxBO+btLBEMAw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-05T23:13:17.788953Z","bundle_sha256":"b41dd0b9d9840cd15a085084235b81ccbbd4dcdd44e9c800bdc93d6481aa30c1"}}