{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:IPAPR5SI2GM2PUH3QRMMXHLPPC","short_pith_number":"pith:IPAPR5SI","schema_version":"1.0","canonical_sha256":"43c0f8f648d199a7d0fb8458cb9d6f788d6334f7897a26122b63d7309cc798b0","source":{"kind":"arxiv","id":"2109.00544","version":2},"attestation_state":"computed","paper":{"title":"Towards Improving Adversarial Training of NLP Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Jin Yong Yoo, Yanjun Qi","submitted_at":"2021-09-01T17:14:26Z","abstract_excerpt":"Adversarial training, a method for learning robust deep neural networks, constructs adversarial examples during training. However, recent methods for generating NLP adversarial examples involve combinatorial search and expensive sentence encoders for constraining the generated instances. As a result, it remains challenging to use vanilla adversarial training to improve NLP models' performance, and the benefits are mainly uninvestigated. This paper proposes a simple and improved vanilla adversarial training process for NLP models, which we name Attacking to Training (A2T). The core part of A2T "},"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":"2109.00544","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2021-09-01T17:14:26Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"0e9ba0e7b236b7f26f20069454888c27ba8be836148a297f71e7cb6426f1e57b","abstract_canon_sha256":"53fb0beb8ba749ea1901279c48e5130d8c5e8d113e8c8896c2c8ebcad96f93a6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:13:32.610767Z","signature_b64":"fiKbqJ/WLkUOCZvVw7xyIQKibp9+3f/SkIXZQalOvDdPNhIEYV5KTplmdWk6ZF+gV7/8cJOnhF3fcjp5cKNrCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"43c0f8f648d199a7d0fb8458cb9d6f788d6334f7897a26122b63d7309cc798b0","last_reissued_at":"2026-07-05T03:13:32.610231Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:13:32.610231Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Towards Improving Adversarial Training of NLP Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Jin Yong Yoo, Yanjun Qi","submitted_at":"2021-09-01T17:14:26Z","abstract_excerpt":"Adversarial training, a method for learning robust deep neural networks, constructs adversarial examples during training. However, recent methods for generating NLP adversarial examples involve combinatorial search and expensive sentence encoders for constraining the generated instances. As a result, it remains challenging to use vanilla adversarial training to improve NLP models' performance, and the benefits are mainly uninvestigated. This paper proposes a simple and improved vanilla adversarial training process for NLP models, which we name Attacking to Training (A2T). The core part of A2T "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2109.00544","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/2109.00544/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":"2109.00544","created_at":"2026-07-05T03:13:32.610297+00:00"},{"alias_kind":"arxiv_version","alias_value":"2109.00544v2","created_at":"2026-07-05T03:13:32.610297+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2109.00544","created_at":"2026-07-05T03:13:32.610297+00:00"},{"alias_kind":"pith_short_12","alias_value":"IPAPR5SI2GM2","created_at":"2026-07-05T03:13:32.610297+00:00"},{"alias_kind":"pith_short_16","alias_value":"IPAPR5SI2GM2PUH3","created_at":"2026-07-05T03:13:32.610297+00:00"},{"alias_kind":"pith_short_8","alias_value":"IPAPR5SI","created_at":"2026-07-05T03:13:32.610297+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.14163","citing_title":"SeaAlert: Robust Severity Classification and LLM-Based Information Extraction for Noisy Maritime Distress Communications","ref_index":23,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IPAPR5SI2GM2PUH3QRMMXHLPPC","json":"https://pith.science/pith/IPAPR5SI2GM2PUH3QRMMXHLPPC.json","graph_json":"https://pith.science/api/pith-number/IPAPR5SI2GM2PUH3QRMMXHLPPC/graph.json","events_json":"https://pith.science/api/pith-number/IPAPR5SI2GM2PUH3QRMMXHLPPC/events.json","paper":"https://pith.science/paper/IPAPR5SI"},"agent_actions":{"view_html":"https://pith.science/pith/IPAPR5SI2GM2PUH3QRMMXHLPPC","download_json":"https://pith.science/pith/IPAPR5SI2GM2PUH3QRMMXHLPPC.json","view_paper":"https://pith.science/paper/IPAPR5SI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2109.00544&json=true","fetch_graph":"https://pith.science/api/pith-number/IPAPR5SI2GM2PUH3QRMMXHLPPC/graph.json","fetch_events":"https://pith.science/api/pith-number/IPAPR5SI2GM2PUH3QRMMXHLPPC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IPAPR5SI2GM2PUH3QRMMXHLPPC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IPAPR5SI2GM2PUH3QRMMXHLPPC/action/storage_attestation","attest_author":"https://pith.science/pith/IPAPR5SI2GM2PUH3QRMMXHLPPC/action/author_attestation","sign_citation":"https://pith.science/pith/IPAPR5SI2GM2PUH3QRMMXHLPPC/action/citation_signature","submit_replication":"https://pith.science/pith/IPAPR5SI2GM2PUH3QRMMXHLPPC/action/replication_record"}},"created_at":"2026-07-05T03:13:32.610297+00:00","updated_at":"2026-07-05T03:13:32.610297+00:00"}