{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:DULWWLFEN7TSPDSNOWKXMCHAXE","short_pith_number":"pith:DULWWLFE","schema_version":"1.0","canonical_sha256":"1d176b2ca46fe7278e4d75957608e0b914badd2789fba3e92db275970526a03b","source":{"kind":"arxiv","id":"2506.05689","version":1},"attestation_state":"computed","paper":{"title":"Pts3D-LLM: Studying the Impact of Token Structure for 3D Scene Understanding With Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Chen Chen, Hugues Thomas, Jian Zhang","submitted_at":"2025-06-06T02:35:26Z","abstract_excerpt":"Effectively representing 3D scenes for Multimodal Large Language Models (MLLMs) is crucial yet challenging. Existing approaches commonly only rely on 2D image features and use varied tokenization approaches. This work presents a rigorous study of 3D token structures, systematically comparing video-based and point-based representations while maintaining consistent model backbones and parameters. We propose a novel approach that enriches visual tokens by incorporating 3D point cloud features from a Sonata pretrained Point Transformer V3 encoder. Our experiments demonstrate that merging explicit "},"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":"2506.05689","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-06-06T02:35:26Z","cross_cats_sorted":[],"title_canon_sha256":"1771f4a156a5cc85b0f76e6489debd64de9a703199ea01130ff7d03739c52658","abstract_canon_sha256":"aa1655a76c789d49c73bd7cd0a73edae1d8d9786b0f5db0165e1749621e03804"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:17:08.107014Z","signature_b64":"1dw5vGihzfyWp+Liokl//vuKFvfSOzJFQe+T6z7c8dPfgnwwZNXp8iANSzHiEjhmRFjQKM/yfUyWE3zH7MwHBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1d176b2ca46fe7278e4d75957608e0b914badd2789fba3e92db275970526a03b","last_reissued_at":"2026-07-05T11:17:08.106511Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:17:08.106511Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Pts3D-LLM: Studying the Impact of Token Structure for 3D Scene Understanding With Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Chen Chen, Hugues Thomas, Jian Zhang","submitted_at":"2025-06-06T02:35:26Z","abstract_excerpt":"Effectively representing 3D scenes for Multimodal Large Language Models (MLLMs) is crucial yet challenging. Existing approaches commonly only rely on 2D image features and use varied tokenization approaches. This work presents a rigorous study of 3D token structures, systematically comparing video-based and point-based representations while maintaining consistent model backbones and parameters. We propose a novel approach that enriches visual tokens by incorporating 3D point cloud features from a Sonata pretrained Point Transformer V3 encoder. Our experiments demonstrate that merging explicit "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.05689","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/2506.05689/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":"2506.05689","created_at":"2026-07-05T11:17:08.106571+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.05689v1","created_at":"2026-07-05T11:17:08.106571+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.05689","created_at":"2026-07-05T11:17:08.106571+00:00"},{"alias_kind":"pith_short_12","alias_value":"DULWWLFEN7TS","created_at":"2026-07-05T11:17:08.106571+00:00"},{"alias_kind":"pith_short_16","alias_value":"DULWWLFEN7TSPDSN","created_at":"2026-07-05T11:17:08.106571+00:00"},{"alias_kind":"pith_short_8","alias_value":"DULWWLFE","created_at":"2026-07-05T11:17:08.106571+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.06534","citing_title":"CAIRN: Cross-Room 3D Scene Understanding with Topology-Aware Large Multimodal Models","ref_index":48,"is_internal_anchor":true},{"citing_arxiv_id":"2606.19828","citing_title":"3D-PLOT-LLM: Part-Level Object Tokens for 3D Large Language Models","ref_index":5,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/DULWWLFEN7TSPDSNOWKXMCHAXE","json":"https://pith.science/pith/DULWWLFEN7TSPDSNOWKXMCHAXE.json","graph_json":"https://pith.science/api/pith-number/DULWWLFEN7TSPDSNOWKXMCHAXE/graph.json","events_json":"https://pith.science/api/pith-number/DULWWLFEN7TSPDSNOWKXMCHAXE/events.json","paper":"https://pith.science/paper/DULWWLFE"},"agent_actions":{"view_html":"https://pith.science/pith/DULWWLFEN7TSPDSNOWKXMCHAXE","download_json":"https://pith.science/pith/DULWWLFEN7TSPDSNOWKXMCHAXE.json","view_paper":"https://pith.science/paper/DULWWLFE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.05689&json=true","fetch_graph":"https://pith.science/api/pith-number/DULWWLFEN7TSPDSNOWKXMCHAXE/graph.json","fetch_events":"https://pith.science/api/pith-number/DULWWLFEN7TSPDSNOWKXMCHAXE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DULWWLFEN7TSPDSNOWKXMCHAXE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DULWWLFEN7TSPDSNOWKXMCHAXE/action/storage_attestation","attest_author":"https://pith.science/pith/DULWWLFEN7TSPDSNOWKXMCHAXE/action/author_attestation","sign_citation":"https://pith.science/pith/DULWWLFEN7TSPDSNOWKXMCHAXE/action/citation_signature","submit_replication":"https://pith.science/pith/DULWWLFEN7TSPDSNOWKXMCHAXE/action/replication_record"}},"created_at":"2026-07-05T11:17:08.106571+00:00","updated_at":"2026-07-05T11:17:08.106571+00:00"}