{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:VYFP46HCMEAIPQ3VGZEDIK7FG3","short_pith_number":"pith:VYFP46HC","schema_version":"1.0","canonical_sha256":"ae0afe78e2610087c3753648342be536c23cc98c893e04adf862d38b5ea53c51","source":{"kind":"arxiv","id":"2103.15543","version":1},"attestation_state":"computed","paper":{"title":"Be Careful about Poisoned Word Embeddings: Exploring the Vulnerability of the Embedding Layers in NLP Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Bin He, Lei Li, Wenkai Yang, Xuancheng Ren, Xu Sun, Zhiyuan Zhang","submitted_at":"2021-03-29T12:19:45Z","abstract_excerpt":"Recent studies have revealed a security threat to natural language processing (NLP) models, called the Backdoor Attack. Victim models can maintain competitive performance on clean samples while behaving abnormally on samples with a specific trigger word inserted. Previous backdoor attacking methods usually assume that attackers have a certain degree of data knowledge, either the dataset which users would use or proxy datasets for a similar task, for implementing the data poisoning procedure. However, in this paper, we find that it is possible to hack the model in a data-free way by modifying o"},"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":"2103.15543","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2021-03-29T12:19:45Z","cross_cats_sorted":[],"title_canon_sha256":"0903a4fa06dfb069fe904d9ef2278ecbaca59983f54c697b362bc1149fcde847","abstract_canon_sha256":"f5cf5c93278255de4a9e1797c94f0738aae0356d171def787fcc3680b48157db"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:27:07.229384Z","signature_b64":"X0j1QC4uhaGNwOrvE6XHKHXI5/mt3kFbEKWUA0gm/Ri1mKlvxOQYQca+amXqwdTB0JuPSwkSNZnjr8CxwvriDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ae0afe78e2610087c3753648342be536c23cc98c893e04adf862d38b5ea53c51","last_reissued_at":"2026-07-05T02:27:07.228917Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:27:07.228917Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Be Careful about Poisoned Word Embeddings: Exploring the Vulnerability of the Embedding Layers in NLP Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Bin He, Lei Li, Wenkai Yang, Xuancheng Ren, Xu Sun, Zhiyuan Zhang","submitted_at":"2021-03-29T12:19:45Z","abstract_excerpt":"Recent studies have revealed a security threat to natural language processing (NLP) models, called the Backdoor Attack. Victim models can maintain competitive performance on clean samples while behaving abnormally on samples with a specific trigger word inserted. Previous backdoor attacking methods usually assume that attackers have a certain degree of data knowledge, either the dataset which users would use or proxy datasets for a similar task, for implementing the data poisoning procedure. However, in this paper, we find that it is possible to hack the model in a data-free way by modifying o"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2103.15543","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/2103.15543/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":"2103.15543","created_at":"2026-07-05T02:27:07.228975+00:00"},{"alias_kind":"arxiv_version","alias_value":"2103.15543v1","created_at":"2026-07-05T02:27:07.228975+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2103.15543","created_at":"2026-07-05T02:27:07.228975+00:00"},{"alias_kind":"pith_short_12","alias_value":"VYFP46HCMEAI","created_at":"2026-07-05T02:27:07.228975+00:00"},{"alias_kind":"pith_short_16","alias_value":"VYFP46HCMEAIPQ3V","created_at":"2026-07-05T02:27:07.228975+00:00"},{"alias_kind":"pith_short_8","alias_value":"VYFP46HC","created_at":"2026-07-05T02:27:07.228975+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.30027","citing_title":"DocRetriever: A Plug-and-Play Framework for Multimodal Document Retrieval with Comprehensive Benchmark","ref_index":69,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10611","citing_title":"Re-Triggering Safeguards within LLMs for Jailbreak Detection","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2604.05480","citing_title":"Can You Trust the Vectors in Your Vector Database? Black-Hole Attack from Embedding Space Defects","ref_index":59,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/VYFP46HCMEAIPQ3VGZEDIK7FG3","json":"https://pith.science/pith/VYFP46HCMEAIPQ3VGZEDIK7FG3.json","graph_json":"https://pith.science/api/pith-number/VYFP46HCMEAIPQ3VGZEDIK7FG3/graph.json","events_json":"https://pith.science/api/pith-number/VYFP46HCMEAIPQ3VGZEDIK7FG3/events.json","paper":"https://pith.science/paper/VYFP46HC"},"agent_actions":{"view_html":"https://pith.science/pith/VYFP46HCMEAIPQ3VGZEDIK7FG3","download_json":"https://pith.science/pith/VYFP46HCMEAIPQ3VGZEDIK7FG3.json","view_paper":"https://pith.science/paper/VYFP46HC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2103.15543&json=true","fetch_graph":"https://pith.science/api/pith-number/VYFP46HCMEAIPQ3VGZEDIK7FG3/graph.json","fetch_events":"https://pith.science/api/pith-number/VYFP46HCMEAIPQ3VGZEDIK7FG3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/VYFP46HCMEAIPQ3VGZEDIK7FG3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/VYFP46HCMEAIPQ3VGZEDIK7FG3/action/storage_attestation","attest_author":"https://pith.science/pith/VYFP46HCMEAIPQ3VGZEDIK7FG3/action/author_attestation","sign_citation":"https://pith.science/pith/VYFP46HCMEAIPQ3VGZEDIK7FG3/action/citation_signature","submit_replication":"https://pith.science/pith/VYFP46HCMEAIPQ3VGZEDIK7FG3/action/replication_record"}},"created_at":"2026-07-05T02:27:07.228975+00:00","updated_at":"2026-07-05T02:27:07.228975+00:00"}