{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:FH3XBLA24G3N7F4KLOBPMTI4SW","short_pith_number":"pith:FH3XBLA2","schema_version":"1.0","canonical_sha256":"29f770ac1ae1b6df978a5b82f64d1c9585454351f72d76e961a0e3c13327bb93","source":{"kind":"arxiv","id":"2311.05043","version":1},"attestation_state":"computed","paper":{"title":"Zero-shot Translation of Attention Patterns in VQA Models to Natural Language","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.CV","authors_text":"A. Sophia Koepke, Hendrik P. A. Lensch, Leonard Salewski, Zeynep Akata","submitted_at":"2023-11-08T22:18:53Z","abstract_excerpt":"Converting a model's internals to text can yield human-understandable insights about the model. Inspired by the recent success of training-free approaches for image captioning, we propose ZS-A2T, a zero-shot framework that translates the transformer attention of a given model into natural language without requiring any training. We consider this in the context of Visual Question Answering (VQA). ZS-A2T builds on a pre-trained large language model (LLM), which receives a task prompt, question, and predicted answer, as inputs. The LLM is guided to select tokens which describe the regions in the "},"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":"2311.05043","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2023-11-08T22:18:53Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"d825ea2be6fd6003f1c9718032ddefe7bd898a149cd61a10db64bea82eca70cc","abstract_canon_sha256":"036a04298c9f3068979c7decc97d1dc7f7698754347a5f4e4503e0100afa12b0"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:10:53.882556Z","signature_b64":"kc4ftb6sO7cA2DiMG3Sruu8jMshmIX/lJT/1KfQSBgNMynDpMsU2QUYkMW8sJFCD2Jsx3X92jwCgnS5VaKZ+Aw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"29f770ac1ae1b6df978a5b82f64d1c9585454351f72d76e961a0e3c13327bb93","last_reissued_at":"2026-07-05T07:10:53.882220Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:10:53.882220Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Zero-shot Translation of Attention Patterns in VQA Models to Natural Language","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.CV","authors_text":"A. Sophia Koepke, Hendrik P. A. Lensch, Leonard Salewski, Zeynep Akata","submitted_at":"2023-11-08T22:18:53Z","abstract_excerpt":"Converting a model's internals to text can yield human-understandable insights about the model. Inspired by the recent success of training-free approaches for image captioning, we propose ZS-A2T, a zero-shot framework that translates the transformer attention of a given model into natural language without requiring any training. We consider this in the context of Visual Question Answering (VQA). ZS-A2T builds on a pre-trained large language model (LLM), which receives a task prompt, question, and predicted answer, as inputs. The LLM is guided to select tokens which describe the regions in the "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2311.05043","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/2311.05043/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":"2311.05043","created_at":"2026-07-05T07:10:53.882267+00:00"},{"alias_kind":"arxiv_version","alias_value":"2311.05043v1","created_at":"2026-07-05T07:10:53.882267+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2311.05043","created_at":"2026-07-05T07:10:53.882267+00:00"},{"alias_kind":"pith_short_12","alias_value":"FH3XBLA24G3N","created_at":"2026-07-05T07:10:53.882267+00:00"},{"alias_kind":"pith_short_16","alias_value":"FH3XBLA24G3N7F4K","created_at":"2026-07-05T07:10:53.882267+00:00"},{"alias_kind":"pith_short_8","alias_value":"FH3XBLA2","created_at":"2026-07-05T07:10:53.882267+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FH3XBLA24G3N7F4KLOBPMTI4SW","json":"https://pith.science/pith/FH3XBLA24G3N7F4KLOBPMTI4SW.json","graph_json":"https://pith.science/api/pith-number/FH3XBLA24G3N7F4KLOBPMTI4SW/graph.json","events_json":"https://pith.science/api/pith-number/FH3XBLA24G3N7F4KLOBPMTI4SW/events.json","paper":"https://pith.science/paper/FH3XBLA2"},"agent_actions":{"view_html":"https://pith.science/pith/FH3XBLA24G3N7F4KLOBPMTI4SW","download_json":"https://pith.science/pith/FH3XBLA24G3N7F4KLOBPMTI4SW.json","view_paper":"https://pith.science/paper/FH3XBLA2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2311.05043&json=true","fetch_graph":"https://pith.science/api/pith-number/FH3XBLA24G3N7F4KLOBPMTI4SW/graph.json","fetch_events":"https://pith.science/api/pith-number/FH3XBLA24G3N7F4KLOBPMTI4SW/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FH3XBLA24G3N7F4KLOBPMTI4SW/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FH3XBLA24G3N7F4KLOBPMTI4SW/action/storage_attestation","attest_author":"https://pith.science/pith/FH3XBLA24G3N7F4KLOBPMTI4SW/action/author_attestation","sign_citation":"https://pith.science/pith/FH3XBLA24G3N7F4KLOBPMTI4SW/action/citation_signature","submit_replication":"https://pith.science/pith/FH3XBLA24G3N7F4KLOBPMTI4SW/action/replication_record"}},"created_at":"2026-07-05T07:10:53.882267+00:00","updated_at":"2026-07-05T07:10:53.882267+00:00"}