{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:NU6ITC2BFUFX6RX53J4B4KVCSC","short_pith_number":"pith:NU6ITC2B","schema_version":"1.0","canonical_sha256":"6d3c898b412d0b7f46fdda781e2aa290af4e93515cf29d899e30817d3dc00f53","source":{"kind":"arxiv","id":"2211.03013","version":1},"attestation_state":"computed","paper":{"title":"Robust Lottery Tickets for Pre-trained Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Di Liang, Qi Zhang, Rong Bao, Rui Zheng, Sirui Wang, Tao Gui, Wei Wu, Xuanjing Huang, Yuhao Zhou","submitted_at":"2022-11-06T02:59:27Z","abstract_excerpt":"Recent works on Lottery Ticket Hypothesis have shown that pre-trained language models (PLMs) contain smaller matching subnetworks(winning tickets) which are capable of reaching accuracy comparable to the original models. However, these tickets are proved to be notrobust to adversarial examples, and even worse than their PLM counterparts. To address this problem, we propose a novel method based on learning binary weight masks to identify robust tickets hidden in the original PLMs. Since the loss is not differentiable for the binary mask, we assign the hard concrete distribution to the masks and"},"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":"2211.03013","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2022-11-06T02:59:27Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"d02d6b836fa564ccf18d630afe852b7a00d7c302b9a3a03898373535fdc0bd70","abstract_canon_sha256":"53fb6eb53af5c134d4d928b2398f25b25792bf9ca1117cfd831a3849273754e1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:15:44.160976Z","signature_b64":"iqCxwuNLdECH52bXz2ijJzJSK3fuDSuxX/wBGAWjtBArsJjEoEA+5fSiXaJjFwxe5vuUNH58W1EvySZDQ8SpCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6d3c898b412d0b7f46fdda781e2aa290af4e93515cf29d899e30817d3dc00f53","last_reissued_at":"2026-07-05T05:15:44.160569Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:15:44.160569Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Robust Lottery Tickets for Pre-trained Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Di Liang, Qi Zhang, Rong Bao, Rui Zheng, Sirui Wang, Tao Gui, Wei Wu, Xuanjing Huang, Yuhao Zhou","submitted_at":"2022-11-06T02:59:27Z","abstract_excerpt":"Recent works on Lottery Ticket Hypothesis have shown that pre-trained language models (PLMs) contain smaller matching subnetworks(winning tickets) which are capable of reaching accuracy comparable to the original models. However, these tickets are proved to be notrobust to adversarial examples, and even worse than their PLM counterparts. To address this problem, we propose a novel method based on learning binary weight masks to identify robust tickets hidden in the original PLMs. Since the loss is not differentiable for the binary mask, we assign the hard concrete distribution to the masks and"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2211.03013","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/2211.03013/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":"2211.03013","created_at":"2026-07-05T05:15:44.160625+00:00"},{"alias_kind":"arxiv_version","alias_value":"2211.03013v1","created_at":"2026-07-05T05:15:44.160625+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2211.03013","created_at":"2026-07-05T05:15:44.160625+00:00"},{"alias_kind":"pith_short_12","alias_value":"NU6ITC2BFUFX","created_at":"2026-07-05T05:15:44.160625+00:00"},{"alias_kind":"pith_short_16","alias_value":"NU6ITC2BFUFX6RX5","created_at":"2026-07-05T05:15:44.160625+00:00"},{"alias_kind":"pith_short_8","alias_value":"NU6ITC2B","created_at":"2026-07-05T05:15:44.160625+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.15208","citing_title":"Quantization Undoes Alignment: Bias Emergence in Compressed LLMs Across Models and Precision Levels","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17051","citing_title":"Efficient Task Adaptation in Large Language Models via Selective Parameter Optimization","ref_index":27,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/NU6ITC2BFUFX6RX53J4B4KVCSC","json":"https://pith.science/pith/NU6ITC2BFUFX6RX53J4B4KVCSC.json","graph_json":"https://pith.science/api/pith-number/NU6ITC2BFUFX6RX53J4B4KVCSC/graph.json","events_json":"https://pith.science/api/pith-number/NU6ITC2BFUFX6RX53J4B4KVCSC/events.json","paper":"https://pith.science/paper/NU6ITC2B"},"agent_actions":{"view_html":"https://pith.science/pith/NU6ITC2BFUFX6RX53J4B4KVCSC","download_json":"https://pith.science/pith/NU6ITC2BFUFX6RX53J4B4KVCSC.json","view_paper":"https://pith.science/paper/NU6ITC2B","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2211.03013&json=true","fetch_graph":"https://pith.science/api/pith-number/NU6ITC2BFUFX6RX53J4B4KVCSC/graph.json","fetch_events":"https://pith.science/api/pith-number/NU6ITC2BFUFX6RX53J4B4KVCSC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NU6ITC2BFUFX6RX53J4B4KVCSC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NU6ITC2BFUFX6RX53J4B4KVCSC/action/storage_attestation","attest_author":"https://pith.science/pith/NU6ITC2BFUFX6RX53J4B4KVCSC/action/author_attestation","sign_citation":"https://pith.science/pith/NU6ITC2BFUFX6RX53J4B4KVCSC/action/citation_signature","submit_replication":"https://pith.science/pith/NU6ITC2BFUFX6RX53J4B4KVCSC/action/replication_record"}},"created_at":"2026-07-05T05:15:44.160625+00:00","updated_at":"2026-07-05T05:15:44.160625+00:00"}