{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:PJRJJ2MM2BBLFHNWCKNWOUU3UU","short_pith_number":"pith:PJRJJ2MM","schema_version":"1.0","canonical_sha256":"7a6294e98cd042b29db6129b67529ba50dc569f25a516c5b94c1b729552242da","source":{"kind":"arxiv","id":"2205.11603","version":3},"attestation_state":"computed","paper":{"title":"Representation Projection Invariance Mitigates Representation Collapse","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Anastasia Razdaibiedina, Ashish Khetan, Daniel Khashabi, Vishaal Kapoor, Vivek Madan, Zohar Karnin","submitted_at":"2022-05-23T20:00:22Z","abstract_excerpt":"Fine-tuning contextualized representations learned by pre-trained language models remains a prevalent practice in NLP. However, fine-tuning can lead to representation degradation (also known as representation collapse), which may result in instability, sub-optimal performance, and weak generalization.\n  In this paper, we propose Representation Projection Invariance (REPINA), a novel regularization method to maintain the information content of representation and reduce representation collapse during fine-tuning by discouraging undesirable changes in the representations. We study the empirical b"},"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":"2205.11603","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-05-23T20:00:22Z","cross_cats_sorted":[],"title_canon_sha256":"6a563f25aa95834447dc96eaaac446f9ab193797ed745bfda261dae4627111bd","abstract_canon_sha256":"a7853f30416f3650ae58c6bb218247ef186a4683a7a6fbc4e6faa161a770584c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:15:12.116659Z","signature_b64":"SuGnMP7Q5pRSIArjhKuZPDS4jDTcDgiLVNyQlVhynh01GQg8iXJZIMpwJra9nOpB30B/DEukPPY6QmIhEwWLDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7a6294e98cd042b29db6129b67529ba50dc569f25a516c5b94c1b729552242da","last_reissued_at":"2026-07-05T07:15:12.116230Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:15:12.116230Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Representation Projection Invariance Mitigates Representation Collapse","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Anastasia Razdaibiedina, Ashish Khetan, Daniel Khashabi, Vishaal Kapoor, Vivek Madan, Zohar Karnin","submitted_at":"2022-05-23T20:00:22Z","abstract_excerpt":"Fine-tuning contextualized representations learned by pre-trained language models remains a prevalent practice in NLP. However, fine-tuning can lead to representation degradation (also known as representation collapse), which may result in instability, sub-optimal performance, and weak generalization.\n  In this paper, we propose Representation Projection Invariance (REPINA), a novel regularization method to maintain the information content of representation and reduce representation collapse during fine-tuning by discouraging undesirable changes in the representations. We study the empirical b"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2205.11603","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/2205.11603/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":"2205.11603","created_at":"2026-07-05T07:15:12.116284+00:00"},{"alias_kind":"arxiv_version","alias_value":"2205.11603v3","created_at":"2026-07-05T07:15:12.116284+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2205.11603","created_at":"2026-07-05T07:15:12.116284+00:00"},{"alias_kind":"pith_short_12","alias_value":"PJRJJ2MM2BBL","created_at":"2026-07-05T07:15:12.116284+00:00"},{"alias_kind":"pith_short_16","alias_value":"PJRJJ2MM2BBLFHNW","created_at":"2026-07-05T07:15:12.116284+00:00"},{"alias_kind":"pith_short_8","alias_value":"PJRJJ2MM","created_at":"2026-07-05T07:15:12.116284+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/PJRJJ2MM2BBLFHNWCKNWOUU3UU","json":"https://pith.science/pith/PJRJJ2MM2BBLFHNWCKNWOUU3UU.json","graph_json":"https://pith.science/api/pith-number/PJRJJ2MM2BBLFHNWCKNWOUU3UU/graph.json","events_json":"https://pith.science/api/pith-number/PJRJJ2MM2BBLFHNWCKNWOUU3UU/events.json","paper":"https://pith.science/paper/PJRJJ2MM"},"agent_actions":{"view_html":"https://pith.science/pith/PJRJJ2MM2BBLFHNWCKNWOUU3UU","download_json":"https://pith.science/pith/PJRJJ2MM2BBLFHNWCKNWOUU3UU.json","view_paper":"https://pith.science/paper/PJRJJ2MM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2205.11603&json=true","fetch_graph":"https://pith.science/api/pith-number/PJRJJ2MM2BBLFHNWCKNWOUU3UU/graph.json","fetch_events":"https://pith.science/api/pith-number/PJRJJ2MM2BBLFHNWCKNWOUU3UU/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PJRJJ2MM2BBLFHNWCKNWOUU3UU/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PJRJJ2MM2BBLFHNWCKNWOUU3UU/action/storage_attestation","attest_author":"https://pith.science/pith/PJRJJ2MM2BBLFHNWCKNWOUU3UU/action/author_attestation","sign_citation":"https://pith.science/pith/PJRJJ2MM2BBLFHNWCKNWOUU3UU/action/citation_signature","submit_replication":"https://pith.science/pith/PJRJJ2MM2BBLFHNWCKNWOUU3UU/action/replication_record"}},"created_at":"2026-07-05T07:15:12.116284+00:00","updated_at":"2026-07-05T07:15:12.116284+00:00"}