{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:IMNTB3HPP2GV36C5GMSAOIMRVG","short_pith_number":"pith:IMNTB3HP","schema_version":"1.0","canonical_sha256":"431b30ecef7e8d5df85d3324072191a9bdad6807d365fd7d73a28f5170d54528","source":{"kind":"arxiv","id":"2106.12620","version":2},"attestation_state":"computed","paper":{"title":"IA-RED$^2$: Interpretability-Aware Redundancy Reduction for Vision Transformers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Aude Oliva, Bowen Pan, Rameswar Panda, Rogerio Feris, Yifan Jiang, Zhangyang Wang","submitted_at":"2021-06-23T18:29:23Z","abstract_excerpt":"The self-attention-based model, transformer, is recently becoming the leading backbone in the field of computer vision. In spite of the impressive success made by transformers in a variety of vision tasks, it still suffers from heavy computation and intensive memory costs. To address this limitation, this paper presents an Interpretability-Aware REDundancy REDuction framework (IA-RED$^2$). We start by observing a large amount of redundant computation, mainly spent on uncorrelated input patches, and then introduce an interpretable module to dynamically and gracefully drop these redundant patche"},"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":"2106.12620","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2021-06-23T18:29:23Z","cross_cats_sorted":[],"title_canon_sha256":"f1fa36e11df93c544be86672b5e4d2480068e7c3359337ee3144e2c10169ca09","abstract_canon_sha256":"06d66b7c2b2a0c13fc0381b4313916af460e5c829266eb57a468b1d7c1a6638d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T03:26:18.757251Z","signature_b64":"j+gaQvKbD0K+LxS5JBTCr16vZxKpoAeccakPFn3P4zDZ1/9rB1f+9Od40K8jfQDYBiXW4oCVz3exaJzSRPbWAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"431b30ecef7e8d5df85d3324072191a9bdad6807d365fd7d73a28f5170d54528","last_reissued_at":"2026-07-05T03:26:18.756827Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T03:26:18.756827Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"IA-RED$^2$: Interpretability-Aware Redundancy Reduction for Vision Transformers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Aude Oliva, Bowen Pan, Rameswar Panda, Rogerio Feris, Yifan Jiang, Zhangyang Wang","submitted_at":"2021-06-23T18:29:23Z","abstract_excerpt":"The self-attention-based model, transformer, is recently becoming the leading backbone in the field of computer vision. In spite of the impressive success made by transformers in a variety of vision tasks, it still suffers from heavy computation and intensive memory costs. To address this limitation, this paper presents an Interpretability-Aware REDundancy REDuction framework (IA-RED$^2$). We start by observing a large amount of redundant computation, mainly spent on uncorrelated input patches, and then introduce an interpretable module to dynamically and gracefully drop these redundant patche"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2106.12620","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/2106.12620/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":"2106.12620","created_at":"2026-07-05T03:26:18.756888+00:00"},{"alias_kind":"arxiv_version","alias_value":"2106.12620v2","created_at":"2026-07-05T03:26:18.756888+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2106.12620","created_at":"2026-07-05T03:26:18.756888+00:00"},{"alias_kind":"pith_short_12","alias_value":"IMNTB3HPP2GV","created_at":"2026-07-05T03:26:18.756888+00:00"},{"alias_kind":"pith_short_16","alias_value":"IMNTB3HPP2GV36C5","created_at":"2026-07-05T03:26:18.756888+00:00"},{"alias_kind":"pith_short_8","alias_value":"IMNTB3HP","created_at":"2026-07-05T03:26:18.756888+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2504.16691","citing_title":"Rethinking Vision Transformer for Large-Scale Fine-Grained Image Retrieval","ref_index":56,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IMNTB3HPP2GV36C5GMSAOIMRVG","json":"https://pith.science/pith/IMNTB3HPP2GV36C5GMSAOIMRVG.json","graph_json":"https://pith.science/api/pith-number/IMNTB3HPP2GV36C5GMSAOIMRVG/graph.json","events_json":"https://pith.science/api/pith-number/IMNTB3HPP2GV36C5GMSAOIMRVG/events.json","paper":"https://pith.science/paper/IMNTB3HP"},"agent_actions":{"view_html":"https://pith.science/pith/IMNTB3HPP2GV36C5GMSAOIMRVG","download_json":"https://pith.science/pith/IMNTB3HPP2GV36C5GMSAOIMRVG.json","view_paper":"https://pith.science/paper/IMNTB3HP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2106.12620&json=true","fetch_graph":"https://pith.science/api/pith-number/IMNTB3HPP2GV36C5GMSAOIMRVG/graph.json","fetch_events":"https://pith.science/api/pith-number/IMNTB3HPP2GV36C5GMSAOIMRVG/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IMNTB3HPP2GV36C5GMSAOIMRVG/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IMNTB3HPP2GV36C5GMSAOIMRVG/action/storage_attestation","attest_author":"https://pith.science/pith/IMNTB3HPP2GV36C5GMSAOIMRVG/action/author_attestation","sign_citation":"https://pith.science/pith/IMNTB3HPP2GV36C5GMSAOIMRVG/action/citation_signature","submit_replication":"https://pith.science/pith/IMNTB3HPP2GV36C5GMSAOIMRVG/action/replication_record"}},"created_at":"2026-07-05T03:26:18.756888+00:00","updated_at":"2026-07-05T03:26:18.756888+00:00"}