{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:TYXH2YLO22543DEBW2ZBCVDDHS","short_pith_number":"pith:TYXH2YLO","schema_version":"1.0","canonical_sha256":"9e2e7d616ed6bbcd8c81b6b21154633c8aad241d180744578a8d2247d66e7d1d","source":{"kind":"arxiv","id":"2503.21307","version":1},"attestation_state":"computed","paper":{"title":"InternVL-X: Advancing and Accelerating InternVL Series with Efficient Visual Token Compression","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Dongchen Lu, Huo Cao, Jianliang Zeng, Leping Huang, Mao Shu, Yuyao Sun, Zilu Zhang","submitted_at":"2025-03-27T09:31:35Z","abstract_excerpt":"Most multimodal large language models (MLLMs) treat visual tokens as \"a sequence of text\", integrating them with text tokens into a large language model (LLM). However, a great quantity of visual tokens significantly increases the demand for computational resources and time. In this paper, we propose InternVL-X, which outperforms the InternVL model in both performance and efficiency by incorporating three visual token compression methods. First, we propose a novel vision-language projector, PVTC. This component integrates adjacent visual embeddings to form a local query and utilizes the transf"},"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":"2503.21307","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-03-27T09:31:35Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"def3ba2855a3f99513fe8c4c08dfe22c3be2d4975a456aa9d9a7d36b8280ff15","abstract_canon_sha256":"9427d8afa03fe9ca7e57402556e3b2650024c497e6b34164107982e53d7b5038"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:40:22.931181Z","signature_b64":"D9VrXNLykfgTsenyUx6SiLy0gYrEyzUYyhBdx0DimeU2ZYZO+4cLrgKJ/1SW1hR2VzDR7nKz66aiLzNKw1rXCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9e2e7d616ed6bbcd8c81b6b21154633c8aad241d180744578a8d2247d66e7d1d","last_reissued_at":"2026-07-05T10:40:22.930707Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:40:22.930707Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"InternVL-X: Advancing and Accelerating InternVL Series with Efficient Visual Token Compression","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Dongchen Lu, Huo Cao, Jianliang Zeng, Leping Huang, Mao Shu, Yuyao Sun, Zilu Zhang","submitted_at":"2025-03-27T09:31:35Z","abstract_excerpt":"Most multimodal large language models (MLLMs) treat visual tokens as \"a sequence of text\", integrating them with text tokens into a large language model (LLM). However, a great quantity of visual tokens significantly increases the demand for computational resources and time. In this paper, we propose InternVL-X, which outperforms the InternVL model in both performance and efficiency by incorporating three visual token compression methods. First, we propose a novel vision-language projector, PVTC. This component integrates adjacent visual embeddings to form a local query and utilizes the transf"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.21307","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/2503.21307/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":"2503.21307","created_at":"2026-07-05T10:40:22.930766+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.21307v1","created_at":"2026-07-05T10:40:22.930766+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.21307","created_at":"2026-07-05T10:40:22.930766+00:00"},{"alias_kind":"pith_short_12","alias_value":"TYXH2YLO2254","created_at":"2026-07-05T10:40:22.930766+00:00"},{"alias_kind":"pith_short_16","alias_value":"TYXH2YLO22543DEB","created_at":"2026-07-05T10:40:22.930766+00:00"},{"alias_kind":"pith_short_8","alias_value":"TYXH2YLO","created_at":"2026-07-05T10:40:22.930766+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.27161","citing_title":"TOPS: First-Principles Visual Token Pruning via Constructing Token Optimal Preservation Sets for Efficient MLLM Inference","ref_index":83,"is_internal_anchor":false},{"citing_arxiv_id":"2605.28115","citing_title":"CIVIC: End-to-End Sequence Compactness for Efficient Vision-Language Models","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08329","citing_title":"An Efficient Token Compression Framework for Visual Object Tracking","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08985","citing_title":"LLaVA-UHD v4: What Makes Efficient Visual Encoding in MLLMs?","ref_index":28,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/TYXH2YLO22543DEBW2ZBCVDDHS","json":"https://pith.science/pith/TYXH2YLO22543DEBW2ZBCVDDHS.json","graph_json":"https://pith.science/api/pith-number/TYXH2YLO22543DEBW2ZBCVDDHS/graph.json","events_json":"https://pith.science/api/pith-number/TYXH2YLO22543DEBW2ZBCVDDHS/events.json","paper":"https://pith.science/paper/TYXH2YLO"},"agent_actions":{"view_html":"https://pith.science/pith/TYXH2YLO22543DEBW2ZBCVDDHS","download_json":"https://pith.science/pith/TYXH2YLO22543DEBW2ZBCVDDHS.json","view_paper":"https://pith.science/paper/TYXH2YLO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.21307&json=true","fetch_graph":"https://pith.science/api/pith-number/TYXH2YLO22543DEBW2ZBCVDDHS/graph.json","fetch_events":"https://pith.science/api/pith-number/TYXH2YLO22543DEBW2ZBCVDDHS/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TYXH2YLO22543DEBW2ZBCVDDHS/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TYXH2YLO22543DEBW2ZBCVDDHS/action/storage_attestation","attest_author":"https://pith.science/pith/TYXH2YLO22543DEBW2ZBCVDDHS/action/author_attestation","sign_citation":"https://pith.science/pith/TYXH2YLO22543DEBW2ZBCVDDHS/action/citation_signature","submit_replication":"https://pith.science/pith/TYXH2YLO22543DEBW2ZBCVDDHS/action/replication_record"}},"created_at":"2026-07-05T10:40:22.930766+00:00","updated_at":"2026-07-05T10:40:22.930766+00:00"}