{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:6CYWULPK7AY3ROZU6TJIYN6T7M","short_pith_number":"pith:6CYWULPK","schema_version":"1.0","canonical_sha256":"f0b16a2deaf831b8bb34f4d28c37d3fb19097808cd013f2c2ea3b09051448a94","source":{"kind":"arxiv","id":"2511.16449","version":4},"attestation_state":"computed","paper":{"title":"Bridging the Semantic-Action Gap in Visual Token Pruning for Efficient VLA Inference","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Bo Zhao, Hongyi Cai, Shuo Yang, Tao Lin, Yeqiu Chen, Zheng Liu, Ziyan Liu","submitted_at":"2025-11-20T15:16:09Z","abstract_excerpt":"Vision-Language-Action (VLA) models have shown great potential for embodied AI by integrating visual perception, language understanding, and action execution. In real-time deployment, these models must process continuous visual streams, incurring substantial computational overhead. Visual token pruning -- a mainstream technique for accelerating Vision-Language Models (VLMs) by retaining salient tokens while discarding redundant ones -- offers a natural candidate solution to this challenge. However, directly applying VLM-oriented pruning methods to VLA inference can cause severe degradation in "},"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":"2511.16449","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-11-20T15:16:09Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"c78f213dfa130bef43203f447430875a97ca22f7e2ffec92c0479e3f37a64644","abstract_canon_sha256":"3c090edfd38b441b00994a2166db2527b2ebb92134f6730fd4b06166a84c6de2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-26T02:05:04.187459Z","signature_b64":"KnquemUFlItDENjUKFzeJ8D4ftUoxGcI6CBI00zUBhmMA+2xE9+MAC4FpQjqEzGVplOx5UuMZagMbkO5KFzVDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f0b16a2deaf831b8bb34f4d28c37d3fb19097808cd013f2c2ea3b09051448a94","last_reissued_at":"2026-05-26T02:05:04.186472Z","signature_status":"signed_v1","first_computed_at":"2026-05-26T02:05:04.186472Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Bridging the Semantic-Action Gap in Visual Token Pruning for Efficient VLA Inference","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Bo Zhao, Hongyi Cai, Shuo Yang, Tao Lin, Yeqiu Chen, Zheng Liu, Ziyan Liu","submitted_at":"2025-11-20T15:16:09Z","abstract_excerpt":"Vision-Language-Action (VLA) models have shown great potential for embodied AI by integrating visual perception, language understanding, and action execution. In real-time deployment, these models must process continuous visual streams, incurring substantial computational overhead. Visual token pruning -- a mainstream technique for accelerating Vision-Language Models (VLMs) by retaining salient tokens while discarding redundant ones -- offers a natural candidate solution to this challenge. However, directly applying VLM-oriented pruning methods to VLA inference can cause severe degradation in "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2511.16449","kind":"arxiv","version":4},"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/2511.16449/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":"2511.16449","created_at":"2026-05-26T02:05:04.186625+00:00"},{"alias_kind":"arxiv_version","alias_value":"2511.16449v4","created_at":"2026-05-26T02:05:04.186625+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2511.16449","created_at":"2026-05-26T02:05:04.186625+00:00"},{"alias_kind":"pith_short_12","alias_value":"6CYWULPK7AY3","created_at":"2026-05-26T02:05:04.186625+00:00"},{"alias_kind":"pith_short_16","alias_value":"6CYWULPK7AY3ROZU","created_at":"2026-05-26T02:05:04.186625+00:00"},{"alias_kind":"pith_short_8","alias_value":"6CYWULPK","created_at":"2026-05-26T02:05:04.186625+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":4,"sample":[{"citing_arxiv_id":"2603.19199","citing_title":"FASTER: Rethinking Real-Time Flow VLAs","ref_index":55,"is_internal_anchor":true},{"citing_arxiv_id":"2603.19199","citing_title":"FASTER: Rethinking Real-Time Flow VLAs","ref_index":55,"is_internal_anchor":true},{"citing_arxiv_id":"2605.12160","citing_title":"Premover: Fast Vision-Language-Action Control by Acting Before Instructions Are Complete","ref_index":13,"is_internal_anchor":true},{"citing_arxiv_id":"2605.02739","citing_title":"Latent Bridge: Feature Delta Prediction for Efficient Dual-System Vision-Language-Action Model Inference","ref_index":7,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6CYWULPK7AY3ROZU6TJIYN6T7M","json":"https://pith.science/pith/6CYWULPK7AY3ROZU6TJIYN6T7M.json","graph_json":"https://pith.science/api/pith-number/6CYWULPK7AY3ROZU6TJIYN6T7M/graph.json","events_json":"https://pith.science/api/pith-number/6CYWULPK7AY3ROZU6TJIYN6T7M/events.json","paper":"https://pith.science/paper/6CYWULPK"},"agent_actions":{"view_html":"https://pith.science/pith/6CYWULPK7AY3ROZU6TJIYN6T7M","download_json":"https://pith.science/pith/6CYWULPK7AY3ROZU6TJIYN6T7M.json","view_paper":"https://pith.science/paper/6CYWULPK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2511.16449&json=true","fetch_graph":"https://pith.science/api/pith-number/6CYWULPK7AY3ROZU6TJIYN6T7M/graph.json","fetch_events":"https://pith.science/api/pith-number/6CYWULPK7AY3ROZU6TJIYN6T7M/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6CYWULPK7AY3ROZU6TJIYN6T7M/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6CYWULPK7AY3ROZU6TJIYN6T7M/action/storage_attestation","attest_author":"https://pith.science/pith/6CYWULPK7AY3ROZU6TJIYN6T7M/action/author_attestation","sign_citation":"https://pith.science/pith/6CYWULPK7AY3ROZU6TJIYN6T7M/action/citation_signature","submit_replication":"https://pith.science/pith/6CYWULPK7AY3ROZU6TJIYN6T7M/action/replication_record"}},"created_at":"2026-05-26T02:05:04.186625+00:00","updated_at":"2026-05-26T02:05:04.186625+00:00"}