{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:A6AGX5F56T65HMYHNRXFRQHAND","short_pith_number":"pith:A6AGX5F5","schema_version":"1.0","canonical_sha256":"07806bf4bdf4fdd3b3076c6e58c0e068fb4ca3031713a687f05099eca7ad6ea2","source":{"kind":"arxiv","id":"2004.07667","version":2},"attestation_state":"computed","paper":{"title":"Null It Out: Guarding Protected Attributes by Iterative Nullspace Projection","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Hila Gonen, Michael Twiton, Shauli Ravfogel, Yanai Elazar, Yoav Goldberg","submitted_at":"2020-04-16T14:02:50Z","abstract_excerpt":"The ability to control for the kinds of information encoded in neural representation has a variety of use cases, especially in light of the challenge of interpreting these models. We present Iterative Null-space Projection (INLP), a novel method for removing information from neural representations. Our method is based on repeated training of linear classifiers that predict a certain property we aim to remove, followed by projection of the representations on their null-space. By doing so, the classifiers become oblivious to that target property, making it hard to linearly separate the data acco"},"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":"2004.07667","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-04-16T14:02:50Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"b5bf910d7113e0a1f19941ac4c727940a283be5fc780462c1b8806a18e1f735a","abstract_canon_sha256":"a8c518bc1b0afde8f331e28f89549d42f9722ab2c7f6f22ebb00858388af21a1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T00:59:03.401426Z","signature_b64":"ixpA8f6AUncVGABo3KGpQu0CC1N2mqFUC/u7V9ivu4I6i10eOLpQ7kvQqAKWJp2ut6qPFcXJghYcBoGZ0qL/AA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"07806bf4bdf4fdd3b3076c6e58c0e068fb4ca3031713a687f05099eca7ad6ea2","last_reissued_at":"2026-07-05T00:59:03.400952Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T00:59:03.400952Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Null It Out: Guarding Protected Attributes by Iterative Nullspace Projection","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Hila Gonen, Michael Twiton, Shauli Ravfogel, Yanai Elazar, Yoav Goldberg","submitted_at":"2020-04-16T14:02:50Z","abstract_excerpt":"The ability to control for the kinds of information encoded in neural representation has a variety of use cases, especially in light of the challenge of interpreting these models. We present Iterative Null-space Projection (INLP), a novel method for removing information from neural representations. Our method is based on repeated training of linear classifiers that predict a certain property we aim to remove, followed by projection of the representations on their null-space. By doing so, the classifiers become oblivious to that target property, making it hard to linearly separate the data acco"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2004.07667","kind":"arxiv","version":2},"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/2004.07667/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":"2004.07667","created_at":"2026-07-05T00:59:03.401010+00:00"},{"alias_kind":"arxiv_version","alias_value":"2004.07667v2","created_at":"2026-07-05T00:59:03.401010+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2004.07667","created_at":"2026-07-05T00:59:03.401010+00:00"},{"alias_kind":"pith_short_12","alias_value":"A6AGX5F56T65","created_at":"2026-07-05T00:59:03.401010+00:00"},{"alias_kind":"pith_short_16","alias_value":"A6AGX5F56T65HMYH","created_at":"2026-07-05T00:59:03.401010+00:00"},{"alias_kind":"pith_short_8","alias_value":"A6AGX5F5","created_at":"2026-07-05T00:59:03.401010+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.13339","citing_title":"Probing Persona-Dependent Preferences in Language Models","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2506.13727","citing_title":"Attribution-Guided Pruning for Insight and Control: Circuit Discovery and Targeted Correction in Small-scale LLMs","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2406.11717","citing_title":"Refusal in Language Models Is Mediated by a Single Direction","ref_index":176,"is_internal_anchor":false},{"citing_arxiv_id":"2303.08112","citing_title":"Eliciting Latent Predictions from Transformers with the Tuned Lens","ref_index":74,"is_internal_anchor":false},{"citing_arxiv_id":"2604.10627","citing_title":"Computational Lesions in Multilingual Language Models Separate Shared and Language-specific Brain Alignment","ref_index":64,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07631","citing_title":"Inference Time Causal Probing in LLMs","ref_index":15,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/A6AGX5F56T65HMYHNRXFRQHAND","json":"https://pith.science/pith/A6AGX5F56T65HMYHNRXFRQHAND.json","graph_json":"https://pith.science/api/pith-number/A6AGX5F56T65HMYHNRXFRQHAND/graph.json","events_json":"https://pith.science/api/pith-number/A6AGX5F56T65HMYHNRXFRQHAND/events.json","paper":"https://pith.science/paper/A6AGX5F5"},"agent_actions":{"view_html":"https://pith.science/pith/A6AGX5F56T65HMYHNRXFRQHAND","download_json":"https://pith.science/pith/A6AGX5F56T65HMYHNRXFRQHAND.json","view_paper":"https://pith.science/paper/A6AGX5F5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2004.07667&json=true","fetch_graph":"https://pith.science/api/pith-number/A6AGX5F56T65HMYHNRXFRQHAND/graph.json","fetch_events":"https://pith.science/api/pith-number/A6AGX5F56T65HMYHNRXFRQHAND/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/A6AGX5F56T65HMYHNRXFRQHAND/action/timestamp_anchor","attest_storage":"https://pith.science/pith/A6AGX5F56T65HMYHNRXFRQHAND/action/storage_attestation","attest_author":"https://pith.science/pith/A6AGX5F56T65HMYHNRXFRQHAND/action/author_attestation","sign_citation":"https://pith.science/pith/A6AGX5F56T65HMYHNRXFRQHAND/action/citation_signature","submit_replication":"https://pith.science/pith/A6AGX5F56T65HMYHNRXFRQHAND/action/replication_record"}},"created_at":"2026-07-05T00:59:03.401010+00:00","updated_at":"2026-07-05T00:59:03.401010+00:00"}