{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:CLCLBMVMPAY5K7QSEBLFTF6YCE","short_pith_number":"pith:CLCLBMVM","schema_version":"1.0","canonical_sha256":"12c4b0b2ac7831d57e1220565997d8110f5d6e55b76a74ff40a7cce8087161f0","source":{"kind":"arxiv","id":"2302.07937","version":2},"attestation_state":"computed","paper":{"title":"The Expressive Power of Tuning Only the Normalization Layers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Angeliki Giannou, Dimitris Papailiopoulos, Shashank Rajput","submitted_at":"2023-02-15T20:44:31Z","abstract_excerpt":"Feature normalization transforms such as Batch and Layer-Normalization have become indispensable ingredients of state-of-the-art deep neural networks. Recent studies on fine-tuning large pretrained models indicate that just tuning the parameters of these affine transforms can achieve high accuracy for downstream tasks. These findings open the questions about the expressive power of tuning the normalization layers of frozen networks. In this work, we take the first step towards this question and show that for random ReLU networks, fine-tuning only its normalization layers can reconstruct any ta"},"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":"2302.07937","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-02-15T20:44:31Z","cross_cats_sorted":["cs.AI","stat.ML"],"title_canon_sha256":"5405d2cf608606064dfa994ae4c3257fb76c582f352f1bbeab490d82ea183f4c","abstract_canon_sha256":"ee33f8d8e6a2cf4590240e20c82a89d4c11c3c0ba4c1f4dbc7935bc8c6b8e34b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:27:46.070347Z","signature_b64":"byr+zvQC5FDmxrKpN0qwR+XPsYtPqXHK532YiMkzO1GW9ebKoHQppSxeWTYHQY/f1OCabmNuX9gTtLowoG/3Cg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"12c4b0b2ac7831d57e1220565997d8110f5d6e55b76a74ff40a7cce8087161f0","last_reissued_at":"2026-07-05T06:27:46.069846Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:27:46.069846Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The Expressive Power of Tuning Only the Normalization Layers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","stat.ML"],"primary_cat":"cs.LG","authors_text":"Angeliki Giannou, Dimitris Papailiopoulos, Shashank Rajput","submitted_at":"2023-02-15T20:44:31Z","abstract_excerpt":"Feature normalization transforms such as Batch and Layer-Normalization have become indispensable ingredients of state-of-the-art deep neural networks. Recent studies on fine-tuning large pretrained models indicate that just tuning the parameters of these affine transforms can achieve high accuracy for downstream tasks. These findings open the questions about the expressive power of tuning the normalization layers of frozen networks. In this work, we take the first step towards this question and show that for random ReLU networks, fine-tuning only its normalization layers can reconstruct any ta"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2302.07937","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/2302.07937/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":"2302.07937","created_at":"2026-07-05T06:27:46.069924+00:00"},{"alias_kind":"arxiv_version","alias_value":"2302.07937v2","created_at":"2026-07-05T06:27:46.069924+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2302.07937","created_at":"2026-07-05T06:27:46.069924+00:00"},{"alias_kind":"pith_short_12","alias_value":"CLCLBMVMPAY5","created_at":"2026-07-05T06:27:46.069924+00:00"},{"alias_kind":"pith_short_16","alias_value":"CLCLBMVMPAY5K7QS","created_at":"2026-07-05T06:27:46.069924+00:00"},{"alias_kind":"pith_short_8","alias_value":"CLCLBMVM","created_at":"2026-07-05T06:27:46.069924+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2607.00338","citing_title":"DroneFINE: Domain-Aware Parameter-Efficient Fine-Tuning of Vision-Language Detectors for Drone Images","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2508.06248","citing_title":"Deepfake Detection that Generalizes Across Benchmarks","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05659","citing_title":"Structural Correspondence and Universal Approximation in Diagonal plus Low-Rank Neural Networks","ref_index":19,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/CLCLBMVMPAY5K7QSEBLFTF6YCE","json":"https://pith.science/pith/CLCLBMVMPAY5K7QSEBLFTF6YCE.json","graph_json":"https://pith.science/api/pith-number/CLCLBMVMPAY5K7QSEBLFTF6YCE/graph.json","events_json":"https://pith.science/api/pith-number/CLCLBMVMPAY5K7QSEBLFTF6YCE/events.json","paper":"https://pith.science/paper/CLCLBMVM"},"agent_actions":{"view_html":"https://pith.science/pith/CLCLBMVMPAY5K7QSEBLFTF6YCE","download_json":"https://pith.science/pith/CLCLBMVMPAY5K7QSEBLFTF6YCE.json","view_paper":"https://pith.science/paper/CLCLBMVM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2302.07937&json=true","fetch_graph":"https://pith.science/api/pith-number/CLCLBMVMPAY5K7QSEBLFTF6YCE/graph.json","fetch_events":"https://pith.science/api/pith-number/CLCLBMVMPAY5K7QSEBLFTF6YCE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CLCLBMVMPAY5K7QSEBLFTF6YCE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CLCLBMVMPAY5K7QSEBLFTF6YCE/action/storage_attestation","attest_author":"https://pith.science/pith/CLCLBMVMPAY5K7QSEBLFTF6YCE/action/author_attestation","sign_citation":"https://pith.science/pith/CLCLBMVMPAY5K7QSEBLFTF6YCE/action/citation_signature","submit_replication":"https://pith.science/pith/CLCLBMVMPAY5K7QSEBLFTF6YCE/action/replication_record"}},"created_at":"2026-07-05T06:27:46.069924+00:00","updated_at":"2026-07-05T06:27:46.069924+00:00"}