{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:WQCOF3POU3IM6CC76W27POBT7W","short_pith_number":"pith:WQCOF3PO","schema_version":"1.0","canonical_sha256":"b404e2edeea6d0cf085ff5b5f7b833fdbab3b2b7f9f278f2dc290b48f33c2759","source":{"kind":"arxiv","id":"2203.12942","version":1},"attestation_state":"computed","paper":{"title":"Generating Data to Mitigate Spurious Correlations in Natural Language Inference Datasets","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CY"],"primary_cat":"cs.CL","authors_text":"Matt Gardner, Pontus Stenetorp, Pradeep Dasigi, Yuxiang WU","submitted_at":"2022-03-24T09:08:05Z","abstract_excerpt":"Natural language processing models often exploit spurious correlations between task-independent features and labels in datasets to perform well only within the distributions they are trained on, while not generalising to different task distributions. We propose to tackle this problem by generating a debiased version of a dataset, which can then be used to train a debiased, off-the-shelf model, by simply replacing its training data. Our approach consists of 1) a method for training data generators to generate high-quality, label-consistent data samples; and 2) a filtering mechanism for removing"},"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":"2203.12942","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2022-03-24T09:08:05Z","cross_cats_sorted":["cs.AI","cs.CY"],"title_canon_sha256":"79cf29f3e863528e27d3b6820962d78cb479c7163f2ea82811b4cc8b091427fa","abstract_canon_sha256":"4054f7b843b4fba5eb4d4c4c656130306782f92183e875f22675fa65bc9119fb"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T04:08:19.150100Z","signature_b64":"Dts6axbZS5i/YOkc/vAmJ+qlwkAaKsBmzg4JEM4HuOhAhjiRHb7KugMAwPq/AVfOlEbuAjKRp4LFuAURkwSEBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b404e2edeea6d0cf085ff5b5f7b833fdbab3b2b7f9f278f2dc290b48f33c2759","last_reissued_at":"2026-07-05T04:08:19.149692Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T04:08:19.149692Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Generating Data to Mitigate Spurious Correlations in Natural Language Inference Datasets","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CY"],"primary_cat":"cs.CL","authors_text":"Matt Gardner, Pontus Stenetorp, Pradeep Dasigi, Yuxiang WU","submitted_at":"2022-03-24T09:08:05Z","abstract_excerpt":"Natural language processing models often exploit spurious correlations between task-independent features and labels in datasets to perform well only within the distributions they are trained on, while not generalising to different task distributions. We propose to tackle this problem by generating a debiased version of a dataset, which can then be used to train a debiased, off-the-shelf model, by simply replacing its training data. Our approach consists of 1) a method for training data generators to generate high-quality, label-consistent data samples; and 2) a filtering mechanism for removing"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2203.12942","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/2203.12942/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":"2203.12942","created_at":"2026-07-05T04:08:19.149750+00:00"},{"alias_kind":"arxiv_version","alias_value":"2203.12942v1","created_at":"2026-07-05T04:08:19.149750+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2203.12942","created_at":"2026-07-05T04:08:19.149750+00:00"},{"alias_kind":"pith_short_12","alias_value":"WQCOF3POU3IM","created_at":"2026-07-05T04:08:19.149750+00:00"},{"alias_kind":"pith_short_16","alias_value":"WQCOF3POU3IM6CC7","created_at":"2026-07-05T04:08:19.149750+00:00"},{"alias_kind":"pith_short_8","alias_value":"WQCOF3PO","created_at":"2026-07-05T04:08:19.149750+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.07596","citing_title":"Shortcuts in the Tail: Debiasing via Post-Hoc Spectral Compression of Fine-Tuning Updates","ref_index":15,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/WQCOF3POU3IM6CC76W27POBT7W","json":"https://pith.science/pith/WQCOF3POU3IM6CC76W27POBT7W.json","graph_json":"https://pith.science/api/pith-number/WQCOF3POU3IM6CC76W27POBT7W/graph.json","events_json":"https://pith.science/api/pith-number/WQCOF3POU3IM6CC76W27POBT7W/events.json","paper":"https://pith.science/paper/WQCOF3PO"},"agent_actions":{"view_html":"https://pith.science/pith/WQCOF3POU3IM6CC76W27POBT7W","download_json":"https://pith.science/pith/WQCOF3POU3IM6CC76W27POBT7W.json","view_paper":"https://pith.science/paper/WQCOF3PO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2203.12942&json=true","fetch_graph":"https://pith.science/api/pith-number/WQCOF3POU3IM6CC76W27POBT7W/graph.json","fetch_events":"https://pith.science/api/pith-number/WQCOF3POU3IM6CC76W27POBT7W/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WQCOF3POU3IM6CC76W27POBT7W/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WQCOF3POU3IM6CC76W27POBT7W/action/storage_attestation","attest_author":"https://pith.science/pith/WQCOF3POU3IM6CC76W27POBT7W/action/author_attestation","sign_citation":"https://pith.science/pith/WQCOF3POU3IM6CC76W27POBT7W/action/citation_signature","submit_replication":"https://pith.science/pith/WQCOF3POU3IM6CC76W27POBT7W/action/replication_record"}},"created_at":"2026-07-05T04:08:19.149750+00:00","updated_at":"2026-07-05T04:08:19.149750+00:00"}