{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:F4Q75CTYZXA5S22QLHXDNKML4P","short_pith_number":"pith:F4Q75CTY","schema_version":"1.0","canonical_sha256":"2f21fe8a78cdc1d96b5059ee36a98be3eb7deaf5cf01b9f17d15e41e1b017f53","source":{"kind":"arxiv","id":"2402.05602","version":2},"attestation_state":"computed","paper":{"title":"AttnLRP: Attention-Aware Layer-Wise Relevance Propagation for Transformers","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.CV","cs.LG"],"primary_cat":"cs.CL","authors_text":"Aakriti Jain, Maximilian Dreyer, Reduan Achtibat, Sayed Mohammad Vakilzadeh Hatefi, Sebastian Lapuschkin, Thomas Wiegand, Wojciech Samek","submitted_at":"2024-02-08T12:01:24Z","abstract_excerpt":"Large Language Models are prone to biased predictions and hallucinations, underlining the paramount importance of understanding their model-internal reasoning process. However, achieving faithful attributions for the entirety of a black-box transformer model and maintaining computational efficiency is an unsolved challenge. By extending the Layer-wise Relevance Propagation attribution method to handle attention layers, we address these challenges effectively. While partial solutions exist, our method is the first to faithfully and holistically attribute not only input but also latent represent"},"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":"2402.05602","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2024-02-08T12:01:24Z","cross_cats_sorted":["cs.AI","cs.CV","cs.LG"],"title_canon_sha256":"cab21d707bfda99406e7e05163cacbed4bcf5dfa8a991cf593745c58865422c1","abstract_canon_sha256":"e9e0aec38ecc22059dabdeb16b1826da76c534b2a78b1ad7a95c46d511fa6215"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:29:21.885723Z","signature_b64":"X6uMKwb+hJ/J9IvFTbdksJZmdX7oF7/zQIc0SHRMAQiLV1OK2FSuK7pVQfofqeE4epkbhQjrqQKzokyiKL4VDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2f21fe8a78cdc1d96b5059ee36a98be3eb7deaf5cf01b9f17d15e41e1b017f53","last_reissued_at":"2026-07-05T08:29:21.885220Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:29:21.885220Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"AttnLRP: Attention-Aware Layer-Wise Relevance Propagation for Transformers","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.CV","cs.LG"],"primary_cat":"cs.CL","authors_text":"Aakriti Jain, Maximilian Dreyer, Reduan Achtibat, Sayed Mohammad Vakilzadeh Hatefi, Sebastian Lapuschkin, Thomas Wiegand, Wojciech Samek","submitted_at":"2024-02-08T12:01:24Z","abstract_excerpt":"Large Language Models are prone to biased predictions and hallucinations, underlining the paramount importance of understanding their model-internal reasoning process. However, achieving faithful attributions for the entirety of a black-box transformer model and maintaining computational efficiency is an unsolved challenge. By extending the Layer-wise Relevance Propagation attribution method to handle attention layers, we address these challenges effectively. While partial solutions exist, our method is the first to faithfully and holistically attribute not only input but also latent represent"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.05602","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/2402.05602/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":"2402.05602","created_at":"2026-07-05T08:29:21.885280+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.05602v2","created_at":"2026-07-05T08:29:21.885280+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.05602","created_at":"2026-07-05T08:29:21.885280+00:00"},{"alias_kind":"pith_short_12","alias_value":"F4Q75CTYZXA5","created_at":"2026-07-05T08:29:21.885280+00:00"},{"alias_kind":"pith_short_16","alias_value":"F4Q75CTYZXA5S22Q","created_at":"2026-07-05T08:29:21.885280+00:00"},{"alias_kind":"pith_short_8","alias_value":"F4Q75CTY","created_at":"2026-07-05T08:29:21.885280+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2602.16608","citing_title":"Explainable AI: Context-Aware Layer-Wise Integrated Gradients for Explaining Transformer Models","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05285","citing_title":"Attribution-Guided Continual Learning for Large Language Models","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2604.14325","citing_title":"Faithfulness Serum: Mitigating the Faithfulness Gap in Textual Explanations of LLM Decisions via Attribution Guidance","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2604.04500","citing_title":"Saliency-R1: Enforcing Interpretable and Faithful Vision-language Reasoning via Saliency-map Alignment Reward","ref_index":2,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/F4Q75CTYZXA5S22QLHXDNKML4P","json":"https://pith.science/pith/F4Q75CTYZXA5S22QLHXDNKML4P.json","graph_json":"https://pith.science/api/pith-number/F4Q75CTYZXA5S22QLHXDNKML4P/graph.json","events_json":"https://pith.science/api/pith-number/F4Q75CTYZXA5S22QLHXDNKML4P/events.json","paper":"https://pith.science/paper/F4Q75CTY"},"agent_actions":{"view_html":"https://pith.science/pith/F4Q75CTYZXA5S22QLHXDNKML4P","download_json":"https://pith.science/pith/F4Q75CTYZXA5S22QLHXDNKML4P.json","view_paper":"https://pith.science/paper/F4Q75CTY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.05602&json=true","fetch_graph":"https://pith.science/api/pith-number/F4Q75CTYZXA5S22QLHXDNKML4P/graph.json","fetch_events":"https://pith.science/api/pith-number/F4Q75CTYZXA5S22QLHXDNKML4P/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/F4Q75CTYZXA5S22QLHXDNKML4P/action/timestamp_anchor","attest_storage":"https://pith.science/pith/F4Q75CTYZXA5S22QLHXDNKML4P/action/storage_attestation","attest_author":"https://pith.science/pith/F4Q75CTYZXA5S22QLHXDNKML4P/action/author_attestation","sign_citation":"https://pith.science/pith/F4Q75CTYZXA5S22QLHXDNKML4P/action/citation_signature","submit_replication":"https://pith.science/pith/F4Q75CTYZXA5S22QLHXDNKML4P/action/replication_record"}},"created_at":"2026-07-05T08:29:21.885280+00:00","updated_at":"2026-07-05T08:29:21.885280+00:00"}