{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:UIMJ67Z3ZSCYM6MUZ77IWG5PVN","short_pith_number":"pith:UIMJ67Z3","schema_version":"1.0","canonical_sha256":"a2189f7f3bcc85867994cffe8b1bafab52a65f1ec77fe1dd5100ce80212a83e9","source":{"kind":"arxiv","id":"2104.13733","version":1},"attestation_state":"computed","paper":{"title":"Gradient-based Adversarial Attacks against Text Transformers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CR","cs.LG"],"primary_cat":"cs.CL","authors_text":"Alexandre Sablayrolles, Chuan Guo, Douwe Kiela, Herv\\'e J\\'egou","submitted_at":"2021-04-15T17:43:43Z","abstract_excerpt":"We propose the first general-purpose gradient-based attack against transformer models. Instead of searching for a single adversarial example, we search for a distribution of adversarial examples parameterized by a continuous-valued matrix, hence enabling gradient-based optimization. We empirically demonstrate that our white-box attack attains state-of-the-art attack performance on a variety of natural language tasks. Furthermore, we show that a powerful black-box transfer attack, enabled by sampling from the adversarial distribution, matches or exceeds existing methods, while only requiring ha"},"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":"2104.13733","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2021-04-15T17:43:43Z","cross_cats_sorted":["cs.AI","cs.CR","cs.LG"],"title_canon_sha256":"9caea6f81847c650901d17b0b7a91e92f5e27a320565fcc69c43e2050792db4d","abstract_canon_sha256":"305e18dbc5fa763ab47ed46aef9347bc5fbe46defdf663e6f6dbfe52b9ba7796"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:35:57.800384Z","signature_b64":"VQJdRlkDJaNZyhr032bV9MG2OaSPHdh1KGelo6/UxA2OsljGmdTJ7Id31Fg/NVjkQaLw7L8PxMJ8t01bvrSVCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a2189f7f3bcc85867994cffe8b1bafab52a65f1ec77fe1dd5100ce80212a83e9","last_reissued_at":"2026-07-05T02:35:57.799956Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:35:57.799956Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Gradient-based Adversarial Attacks against Text Transformers","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CR","cs.LG"],"primary_cat":"cs.CL","authors_text":"Alexandre Sablayrolles, Chuan Guo, Douwe Kiela, Herv\\'e J\\'egou","submitted_at":"2021-04-15T17:43:43Z","abstract_excerpt":"We propose the first general-purpose gradient-based attack against transformer models. Instead of searching for a single adversarial example, we search for a distribution of adversarial examples parameterized by a continuous-valued matrix, hence enabling gradient-based optimization. We empirically demonstrate that our white-box attack attains state-of-the-art attack performance on a variety of natural language tasks. Furthermore, we show that a powerful black-box transfer attack, enabled by sampling from the adversarial distribution, matches or exceeds existing methods, while only requiring ha"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2104.13733","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/2104.13733/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":"2104.13733","created_at":"2026-07-05T02:35:57.800011+00:00"},{"alias_kind":"arxiv_version","alias_value":"2104.13733v1","created_at":"2026-07-05T02:35:57.800011+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2104.13733","created_at":"2026-07-05T02:35:57.800011+00:00"},{"alias_kind":"pith_short_12","alias_value":"UIMJ67Z3ZSCY","created_at":"2026-07-05T02:35:57.800011+00:00"},{"alias_kind":"pith_short_16","alias_value":"UIMJ67Z3ZSCYM6MU","created_at":"2026-07-05T02:35:57.800011+00:00"},{"alias_kind":"pith_short_8","alias_value":"UIMJ67Z3","created_at":"2026-07-05T02:35:57.800011+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":9,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.27215","citing_title":"Vulnerability of Natural Language Classifiers to Evolutionary Generated Adversarial Text","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2606.15531","citing_title":"Greedy Coordinate Diffusion: Effective and Semantically Coherent Adversarial Attacks via Diffusion Guidance","ref_index":51,"is_internal_anchor":false},{"citing_arxiv_id":"2606.11533","citing_title":"AI Researchers Must Help Lead Arms Control to Mitigate Military AI Risks","ref_index":75,"is_internal_anchor":false},{"citing_arxiv_id":"2602.20102","citing_title":"BarrierSteer: LLM Safety via Learning Barrier Steering","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2307.15043","citing_title":"Universal and Transferable Adversarial Attacks on Aligned Language Models","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2210.10760","citing_title":"Scaling Laws for Reward Model Overoptimization","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2309.00614","citing_title":"Baseline Defenses for Adversarial Attacks Against Aligned Language Models","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10582","citing_title":"Guaranteed Jailbreaking Defense via Disrupt-and-Rectify Smoothing","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2604.08499","citing_title":"PIArena: A Platform for Prompt Injection Evaluation","ref_index":5,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/UIMJ67Z3ZSCYM6MUZ77IWG5PVN","json":"https://pith.science/pith/UIMJ67Z3ZSCYM6MUZ77IWG5PVN.json","graph_json":"https://pith.science/api/pith-number/UIMJ67Z3ZSCYM6MUZ77IWG5PVN/graph.json","events_json":"https://pith.science/api/pith-number/UIMJ67Z3ZSCYM6MUZ77IWG5PVN/events.json","paper":"https://pith.science/paper/UIMJ67Z3"},"agent_actions":{"view_html":"https://pith.science/pith/UIMJ67Z3ZSCYM6MUZ77IWG5PVN","download_json":"https://pith.science/pith/UIMJ67Z3ZSCYM6MUZ77IWG5PVN.json","view_paper":"https://pith.science/paper/UIMJ67Z3","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2104.13733&json=true","fetch_graph":"https://pith.science/api/pith-number/UIMJ67Z3ZSCYM6MUZ77IWG5PVN/graph.json","fetch_events":"https://pith.science/api/pith-number/UIMJ67Z3ZSCYM6MUZ77IWG5PVN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UIMJ67Z3ZSCYM6MUZ77IWG5PVN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UIMJ67Z3ZSCYM6MUZ77IWG5PVN/action/storage_attestation","attest_author":"https://pith.science/pith/UIMJ67Z3ZSCYM6MUZ77IWG5PVN/action/author_attestation","sign_citation":"https://pith.science/pith/UIMJ67Z3ZSCYM6MUZ77IWG5PVN/action/citation_signature","submit_replication":"https://pith.science/pith/UIMJ67Z3ZSCYM6MUZ77IWG5PVN/action/replication_record"}},"created_at":"2026-07-05T02:35:57.800011+00:00","updated_at":"2026-07-05T02:35:57.800011+00:00"}