{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:3GXTHBB6T4ZZVX6GALL6ETGP7L","short_pith_number":"pith:3GXTHBB6","schema_version":"1.0","canonical_sha256":"d9af33843e9f339adfc602d7e24ccffac6dc7111cf35a2adf3a6598809076590","source":{"kind":"arxiv","id":"2506.23639","version":1},"attestation_state":"computed","paper":{"title":"Unified Multimodal Understanding via Byte-Pair Visual Encoding","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Hao Luo, Sipeng Zheng, Wanpeng Zhang, Yicheng Feng, Yijiang Li, Zihao Yue, Zongqing Lu","submitted_at":"2025-06-30T09:08:08Z","abstract_excerpt":"Multimodal large language models (MLLMs) have made significant progress in vision-language understanding, yet effectively aligning different modalities remains a fundamental challenge. We present a framework that unifies multimodal understanding by applying byte-pair encoding to visual tokens. Unlike conventional approaches that rely on modality-specific encoders, our method directly incorporates structural information into visual tokens, mirroring successful tokenization strategies in text-only language models. We introduce a priority-guided encoding scheme that considers both frequency and s"},"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":"2506.23639","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-06-30T09:08:08Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"584b9dd86e28f368e897a85232b6bd94c6b29240cb46775bf1860e3752b6e69f","abstract_canon_sha256":"20cbb9ddced41022e7cd0590b481e875f98ef0e3bd99de5a2f6c5ff1aa21e68d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:29:25.510530Z","signature_b64":"t3XFPnwW5IdZXT4p4CQnrEcdTmocHV9KcPrK5b3b8O99u4A+kRgSfbTQBGoVYZM4SEdHYO/FhOassoBxUDUNCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d9af33843e9f339adfc602d7e24ccffac6dc7111cf35a2adf3a6598809076590","last_reissued_at":"2026-07-05T11:29:25.510023Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:29:25.510023Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Unified Multimodal Understanding via Byte-Pair Visual Encoding","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Hao Luo, Sipeng Zheng, Wanpeng Zhang, Yicheng Feng, Yijiang Li, Zihao Yue, Zongqing Lu","submitted_at":"2025-06-30T09:08:08Z","abstract_excerpt":"Multimodal large language models (MLLMs) have made significant progress in vision-language understanding, yet effectively aligning different modalities remains a fundamental challenge. We present a framework that unifies multimodal understanding by applying byte-pair encoding to visual tokens. Unlike conventional approaches that rely on modality-specific encoders, our method directly incorporates structural information into visual tokens, mirroring successful tokenization strategies in text-only language models. We introduce a priority-guided encoding scheme that considers both frequency and s"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.23639","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/2506.23639/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":"2506.23639","created_at":"2026-07-05T11:29:25.510082+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.23639v1","created_at":"2026-07-05T11:29:25.510082+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.23639","created_at":"2026-07-05T11:29:25.510082+00:00"},{"alias_kind":"pith_short_12","alias_value":"3GXTHBB6T4ZZ","created_at":"2026-07-05T11:29:25.510082+00:00"},{"alias_kind":"pith_short_16","alias_value":"3GXTHBB6T4ZZVX6G","created_at":"2026-07-05T11:29:25.510082+00:00"},{"alias_kind":"pith_short_8","alias_value":"3GXTHBB6","created_at":"2026-07-05T11:29:25.510082+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2509.02547","citing_title":"The Landscape of Agentic Reinforcement Learning for LLMs: A Survey","ref_index":218,"is_internal_anchor":false},{"citing_arxiv_id":"2604.03893","citing_title":"FeynmanBench: Benchmarking Multimodal LLMs on Diagrammatic Physics Reasoning","ref_index":56,"is_internal_anchor":false},{"citing_arxiv_id":"2605.00078","citing_title":"Being-H0.7: A Latent World-Action Model from Egocentric Videos","ref_index":31,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/3GXTHBB6T4ZZVX6GALL6ETGP7L","json":"https://pith.science/pith/3GXTHBB6T4ZZVX6GALL6ETGP7L.json","graph_json":"https://pith.science/api/pith-number/3GXTHBB6T4ZZVX6GALL6ETGP7L/graph.json","events_json":"https://pith.science/api/pith-number/3GXTHBB6T4ZZVX6GALL6ETGP7L/events.json","paper":"https://pith.science/paper/3GXTHBB6"},"agent_actions":{"view_html":"https://pith.science/pith/3GXTHBB6T4ZZVX6GALL6ETGP7L","download_json":"https://pith.science/pith/3GXTHBB6T4ZZVX6GALL6ETGP7L.json","view_paper":"https://pith.science/paper/3GXTHBB6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.23639&json=true","fetch_graph":"https://pith.science/api/pith-number/3GXTHBB6T4ZZVX6GALL6ETGP7L/graph.json","fetch_events":"https://pith.science/api/pith-number/3GXTHBB6T4ZZVX6GALL6ETGP7L/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/3GXTHBB6T4ZZVX6GALL6ETGP7L/action/timestamp_anchor","attest_storage":"https://pith.science/pith/3GXTHBB6T4ZZVX6GALL6ETGP7L/action/storage_attestation","attest_author":"https://pith.science/pith/3GXTHBB6T4ZZVX6GALL6ETGP7L/action/author_attestation","sign_citation":"https://pith.science/pith/3GXTHBB6T4ZZVX6GALL6ETGP7L/action/citation_signature","submit_replication":"https://pith.science/pith/3GXTHBB6T4ZZVX6GALL6ETGP7L/action/replication_record"}},"created_at":"2026-07-05T11:29:25.510082+00:00","updated_at":"2026-07-05T11:29:25.510082+00:00"}