{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:2GYQR6KA2MGDOCZZBM44IO6DPB","short_pith_number":"pith:2GYQR6KA","schema_version":"1.0","canonical_sha256":"d1b108f940d30c370b390b39c43bc37874bc570e95f8938e4239c1af3a68c105","source":{"kind":"arxiv","id":"2505.05422","version":2},"attestation_state":"computed","paper":{"title":"TokLIP: Marry Visual Tokens to CLIP for Multimodal Comprehension and Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.CV","authors_text":"Haokun Lin, Qingfu Zhang, Teng Wang, Ying Shan, Ying Wei, Yixiao Ge, Yuying Ge, Zhenan Sun, Zhichao Lu","submitted_at":"2025-05-08T17:12:19Z","abstract_excerpt":"Pioneering token-based works such as Chameleon and Emu3 have established a foundation for multimodal unification but face challenges of high training computational overhead and limited comprehension performance due to a lack of high-level semantics. In this paper, we introduce TokLIP, a visual tokenizer that enhances comprehension by semanticizing vector-quantized (VQ) tokens and incorporating CLIP-level semantics while enabling end-to-end multimodal autoregressive training with standard VQ tokens. TokLIP integrates a low-level discrete VQ tokenizer with a ViT-based token encoder to capture hi"},"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":"2505.05422","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-05-08T17:12:19Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"6028440f4ecf5d73a7aa660c0db1ac0cbd688ec121f4d8e24ea63a4371ddc1e5","abstract_canon_sha256":"36a708e0238aa8c92fa6b20371067186617f25d8c6aaaec0757fbc1dc95175e1"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:54:11.996222Z","signature_b64":"A+AtSAQGVeG1Kwv0wlX2kF1RJ5WDk3TE1c9rJDATj50XxC/6QCrmp5f2rLSIR9fDM8xy3tvTy53uNuX/FYK/Aw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d1b108f940d30c370b390b39c43bc37874bc570e95f8938e4239c1af3a68c105","last_reissued_at":"2026-07-05T11:54:11.995716Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:54:11.995716Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TokLIP: Marry Visual Tokens to CLIP for Multimodal Comprehension and Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.CV","authors_text":"Haokun Lin, Qingfu Zhang, Teng Wang, Ying Shan, Ying Wei, Yixiao Ge, Yuying Ge, Zhenan Sun, Zhichao Lu","submitted_at":"2025-05-08T17:12:19Z","abstract_excerpt":"Pioneering token-based works such as Chameleon and Emu3 have established a foundation for multimodal unification but face challenges of high training computational overhead and limited comprehension performance due to a lack of high-level semantics. In this paper, we introduce TokLIP, a visual tokenizer that enhances comprehension by semanticizing vector-quantized (VQ) tokens and incorporating CLIP-level semantics while enabling end-to-end multimodal autoregressive training with standard VQ tokens. TokLIP integrates a low-level discrete VQ tokenizer with a ViT-based token encoder to capture hi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.05422","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/2505.05422/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":"2505.05422","created_at":"2026-07-05T11:54:11.995790+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.05422v2","created_at":"2026-07-05T11:54:11.995790+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.05422","created_at":"2026-07-05T11:54:11.995790+00:00"},{"alias_kind":"pith_short_12","alias_value":"2GYQR6KA2MGD","created_at":"2026-07-05T11:54:11.995790+00:00"},{"alias_kind":"pith_short_16","alias_value":"2GYQR6KA2MGDOCZZ","created_at":"2026-07-05T11:54:11.995790+00:00"},{"alias_kind":"pith_short_8","alias_value":"2GYQR6KA","created_at":"2026-07-05T11:54:11.995790+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":12,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.24849","citing_title":"IV-CoT: Implicit Visual Chain-of-Thought for Structure-Aware Text-to-Image Generation","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2606.24333","citing_title":"UniTranslator: A Unified Multi-modal Framework for End-to-end In-Image Machine Translation","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2606.23041","citing_title":"SPAR: Semantic-Pixel Self-Alignment and Adaptive Routing for Unified Multimodal Models","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2606.23041","citing_title":"SPAR: Semantic-Pixel Self-Alignment and Adaptive Routing for Unified Multimodal Models","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2606.13289","citing_title":"HYDRA-X: Native Unified Multimodal Models with Holistic Visual Tokenizers","ref_index":297,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03578","citing_title":"Diffusing in the Right Space: A Systematic Study of Latent Diffusability","ref_index":84,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01022","citing_title":"ProductWebGen: Benchmarking Multimodal Product Webpage Generation","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2605.18115","citing_title":"WinTok: A Win-Win Hybrid Tokenizer via Decomposing Visual Understanding and Generation with Transferable Tokens","ref_index":55,"is_internal_anchor":false},{"citing_arxiv_id":"2602.01554","citing_title":"InfoTok: Information-Theoretic Regularization for Capacity-Constrained Shared Visual Tokenization in Unified MLLMs","ref_index":47,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14333","citing_title":"InsightTok: Improving Text and Face Fidelity in Discrete Tokenization for Autoregressive Image Generation","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12500","citing_title":"SenseNova-U1: Unifying Multimodal Understanding and Generation with NEO-unify Architecture","ref_index":76,"is_internal_anchor":false},{"citing_arxiv_id":"2506.15564","citing_title":"Show-o2: Improved Native Unified Multimodal Models","ref_index":68,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2GYQR6KA2MGDOCZZBM44IO6DPB","json":"https://pith.science/pith/2GYQR6KA2MGDOCZZBM44IO6DPB.json","graph_json":"https://pith.science/api/pith-number/2GYQR6KA2MGDOCZZBM44IO6DPB/graph.json","events_json":"https://pith.science/api/pith-number/2GYQR6KA2MGDOCZZBM44IO6DPB/events.json","paper":"https://pith.science/paper/2GYQR6KA"},"agent_actions":{"view_html":"https://pith.science/pith/2GYQR6KA2MGDOCZZBM44IO6DPB","download_json":"https://pith.science/pith/2GYQR6KA2MGDOCZZBM44IO6DPB.json","view_paper":"https://pith.science/paper/2GYQR6KA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.05422&json=true","fetch_graph":"https://pith.science/api/pith-number/2GYQR6KA2MGDOCZZBM44IO6DPB/graph.json","fetch_events":"https://pith.science/api/pith-number/2GYQR6KA2MGDOCZZBM44IO6DPB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2GYQR6KA2MGDOCZZBM44IO6DPB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2GYQR6KA2MGDOCZZBM44IO6DPB/action/storage_attestation","attest_author":"https://pith.science/pith/2GYQR6KA2MGDOCZZBM44IO6DPB/action/author_attestation","sign_citation":"https://pith.science/pith/2GYQR6KA2MGDOCZZBM44IO6DPB/action/citation_signature","submit_replication":"https://pith.science/pith/2GYQR6KA2MGDOCZZBM44IO6DPB/action/replication_record"}},"created_at":"2026-07-05T11:54:11.995790+00:00","updated_at":"2026-07-05T11:54:11.995790+00:00"}