{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:GZC5YTVYZ6LUOOFSHTP6URPM2X","short_pith_number":"pith:GZC5YTVY","schema_version":"1.0","canonical_sha256":"3645dc4eb8cf974738b23cdfea45ecd5e140e687b89e1076091f767cb4b72b74","source":{"kind":"arxiv","id":"2608.01985","version":1},"attestation_state":"computed","paper":{"title":"DiffPrune: differentiable information throttling for token pruning in vision-language models","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Landi He, Lijian Xu, Mingde Yao, Shawn Young","submitted_at":"2026-08-03T09:46:32Z","abstract_excerpt":"Visual token pruning reduces the computational cost of Vision-Language Models (VLMs) by removing redundant visual tokens. The key is to learn a score that measures whether a token is useful. Existing methods typically rely on Gumbel-Softmax to approximate discrete selection during training. Such selectors make the score depend on the behavior of a relaxed pruning operator, not directly on the consequence of information loss. In this paper, we propose DiffPrune, which gives token scores a direct meaning. During training, DiffPrune keeps all tokens and weakens each token's information according "},"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":"2608.01985","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CV","submitted_at":"2026-08-03T09:46:32Z","cross_cats_sorted":[],"title_canon_sha256":"dbec50e55a9269606487f8798171b8b726bdd90e371dacabe6e2949f9d2e12b3","abstract_canon_sha256":"206d8c5ed58add72ec0ec04f989d5756c1f4fd74fbd52da7ee0d53a37b76135a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-04T02:10:26.635338Z","signature_b64":"Y1sV1VJYSv220YluqP12Vh8e9J97OpMOLwJujpJK6bhtzyX64DXwpziVSAqjIV9XRf0CLpEb5WL+gh5EuDcxAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3645dc4eb8cf974738b23cdfea45ecd5e140e687b89e1076091f767cb4b72b74","last_reissued_at":"2026-08-04T02:10:26.633603Z","signature_status":"signed_v1","first_computed_at":"2026-08-04T02:10:26.633603Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"DiffPrune: differentiable information throttling for token pruning in vision-language models","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Landi He, Lijian Xu, Mingde Yao, Shawn Young","submitted_at":"2026-08-03T09:46:32Z","abstract_excerpt":"Visual token pruning reduces the computational cost of Vision-Language Models (VLMs) by removing redundant visual tokens. The key is to learn a score that measures whether a token is useful. Existing methods typically rely on Gumbel-Softmax to approximate discrete selection during training. Such selectors make the score depend on the behavior of a relaxed pruning operator, not directly on the consequence of information loss. In this paper, we propose DiffPrune, which gives token scores a direct meaning. During training, DiffPrune keeps all tokens and weakens each token's information according "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.01985","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/2608.01985/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":"2608.01985","created_at":"2026-08-04T02:10:26.635185+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.01985v1","created_at":"2026-08-04T02:10:26.635185+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.01985","created_at":"2026-08-04T02:10:26.635185+00:00"},{"alias_kind":"pith_short_12","alias_value":"GZC5YTVYZ6LU","created_at":"2026-08-04T02:10:26.635185+00:00"},{"alias_kind":"pith_short_16","alias_value":"GZC5YTVYZ6LUOOFS","created_at":"2026-08-04T02:10:26.635185+00:00"},{"alias_kind":"pith_short_8","alias_value":"GZC5YTVY","created_at":"2026-08-04T02:10:26.635185+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/GZC5YTVYZ6LUOOFSHTP6URPM2X","json":"https://pith.science/pith/GZC5YTVYZ6LUOOFSHTP6URPM2X.json","graph_json":"https://pith.science/api/pith-number/GZC5YTVYZ6LUOOFSHTP6URPM2X/graph.json","events_json":"https://pith.science/api/pith-number/GZC5YTVYZ6LUOOFSHTP6URPM2X/events.json","paper":"https://pith.science/paper/GZC5YTVY"},"agent_actions":{"view_html":"https://pith.science/pith/GZC5YTVYZ6LUOOFSHTP6URPM2X","download_json":"https://pith.science/pith/GZC5YTVYZ6LUOOFSHTP6URPM2X.json","view_paper":"https://pith.science/paper/GZC5YTVY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.01985&json=true","fetch_graph":"https://pith.science/api/pith-number/GZC5YTVYZ6LUOOFSHTP6URPM2X/graph.json","fetch_events":"https://pith.science/api/pith-number/GZC5YTVYZ6LUOOFSHTP6URPM2X/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GZC5YTVYZ6LUOOFSHTP6URPM2X/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GZC5YTVYZ6LUOOFSHTP6URPM2X/action/storage_attestation","attest_author":"https://pith.science/pith/GZC5YTVYZ6LUOOFSHTP6URPM2X/action/author_attestation","sign_citation":"https://pith.science/pith/GZC5YTVYZ6LUOOFSHTP6URPM2X/action/citation_signature","submit_replication":"https://pith.science/pith/GZC5YTVYZ6LUOOFSHTP6URPM2X/action/replication_record"}},"created_at":"2026-08-04T02:10:26.635185+00:00","updated_at":"2026-08-04T02:10:26.635185+00:00"}