{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:J7C24QFTMTG7HIXEQVVB52ELBS","short_pith_number":"pith:J7C24QFT","schema_version":"1.0","canonical_sha256":"4fc5ae40b364cdf3a2e4856a1ee88b0caf5268289f9a5e4dc7e8c7bf466f1a25","source":{"kind":"arxiv","id":"2307.03601","version":5},"attestation_state":"computed","paper":{"title":"GPT4RoI: Instruction Tuning Large Language Model on Region-of-Interest","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Kai Chen, Min Xiao, Peize Sun, Ping Luo, Shilong Zhang, Shoufa Chen, Wenqi Shao, Wenwei Zhang, Yu Liu","submitted_at":"2023-07-07T13:43:44Z","abstract_excerpt":"Visual instruction tuning large language model(LLM) on image-text pairs has achieved general-purpose vision-language abilities. However, the lack of region-text pairs limits their advancements to fine-grained multimodal understanding. In this paper, we propose spatial instruction tuning, which introduces the reference to the region-of-interest(RoI) in the instruction. Before sending to LLM, the reference is replaced by RoI features and interleaved with language embeddings as a sequence. Our model GPT4RoI, trained on 7 region-text pair datasets, brings an unprecedented interactive and conversat"},"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":"2307.03601","kind":"arxiv","version":5},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2023-07-07T13:43:44Z","cross_cats_sorted":[],"title_canon_sha256":"a4e8b4c681e39af46b4bce45a12e3f484d14cb942622811a033fd36576b718f2","abstract_canon_sha256":"e036e3509faf0b6dfcedc2651c54c7b48524320d0c94d206d96ceb2023af2b52"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:20:01.418497Z","signature_b64":"m7KTNPfTSpROjTq62bMb77JS1hgkk/bx7oksSTRP5KRcZPk33O3Dn5rGE878CwhdFzyO6lguWyskyiuGYAxPDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4fc5ae40b364cdf3a2e4856a1ee88b0caf5268289f9a5e4dc7e8c7bf466f1a25","last_reissued_at":"2026-07-05T11:20:01.417942Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:20:01.417942Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"GPT4RoI: Instruction Tuning Large Language Model on Region-of-Interest","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Kai Chen, Min Xiao, Peize Sun, Ping Luo, Shilong Zhang, Shoufa Chen, Wenqi Shao, Wenwei Zhang, Yu Liu","submitted_at":"2023-07-07T13:43:44Z","abstract_excerpt":"Visual instruction tuning large language model(LLM) on image-text pairs has achieved general-purpose vision-language abilities. However, the lack of region-text pairs limits their advancements to fine-grained multimodal understanding. In this paper, we propose spatial instruction tuning, which introduces the reference to the region-of-interest(RoI) in the instruction. Before sending to LLM, the reference is replaced by RoI features and interleaved with language embeddings as a sequence. Our model GPT4RoI, trained on 7 region-text pair datasets, brings an unprecedented interactive and conversat"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2307.03601","kind":"arxiv","version":5},"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/2307.03601/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":"2307.03601","created_at":"2026-07-05T11:20:01.418000+00:00"},{"alias_kind":"arxiv_version","alias_value":"2307.03601v5","created_at":"2026-07-05T11:20:01.418000+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2307.03601","created_at":"2026-07-05T11:20:01.418000+00:00"},{"alias_kind":"pith_short_12","alias_value":"J7C24QFTMTG7","created_at":"2026-07-05T11:20:01.418000+00:00"},{"alias_kind":"pith_short_16","alias_value":"J7C24QFTMTG7HIXE","created_at":"2026-07-05T11:20:01.418000+00:00"},{"alias_kind":"pith_short_8","alias_value":"J7C24QFT","created_at":"2026-07-05T11:20:01.418000+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":15,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.19534","citing_title":"PerceptionDLM: Parallel Region Perception with Multimodal Diffusion Language Models","ref_index":57,"is_internal_anchor":false},{"citing_arxiv_id":"2606.11740","citing_title":"UniReason-Med: A Shared Grounded Reasoning Interface for 2D-to-3D Transfer in Medical VQA","ref_index":108,"is_internal_anchor":false},{"citing_arxiv_id":"2606.28998","citing_title":"Reward-Free Code Alignment from Pretrained or Fine-Tuned LLM: Unpacking the Trade-offs for Code Generation","ref_index":62,"is_internal_anchor":false},{"citing_arxiv_id":"2506.04565","citing_title":"From Standalone LLMs to Integrated Intelligence: A Survey of Compound Al Systems","ref_index":225,"is_internal_anchor":false},{"citing_arxiv_id":"2511.21471","citing_title":"SpatialBench: Benchmarking Multimodal Large Language Models for Spatial Cognition","ref_index":71,"is_internal_anchor":false},{"citing_arxiv_id":"2512.10554","citing_title":"Grounding Everything in Tokens for Multimodal Large Language Models","ref_index":88,"is_internal_anchor":false},{"citing_arxiv_id":"2306.13549","citing_title":"A Survey on Multimodal Large Language Models","ref_index":140,"is_internal_anchor":false},{"citing_arxiv_id":"2312.14238","citing_title":"InternVL: Scaling up Vision Foundation Models and Aligning for Generic Visual-Linguistic Tasks","ref_index":182,"is_internal_anchor":false},{"citing_arxiv_id":"2604.03231","citing_title":"CoME-VL: Scaling Complementary Multi-Encoder Vision-Language Learning","ref_index":77,"is_internal_anchor":false},{"citing_arxiv_id":"2404.16821","citing_title":"How Far Are We to GPT-4V? Closing the Gap to Commercial Multimodal Models with Open-Source Suites","ref_index":139,"is_internal_anchor":false},{"citing_arxiv_id":"2310.03744","citing_title":"Improved Baselines with Visual Instruction Tuning","ref_index":56,"is_internal_anchor":false},{"citing_arxiv_id":"2406.06525","citing_title":"Autoregressive Model Beats Diffusion: Llama for Scalable Image Generation","ref_index":42,"is_internal_anchor":false},{"citing_arxiv_id":"2605.00663","citing_title":"Affordance Agent Harness: Verification-Gated Skill Orchestration","ref_index":84,"is_internal_anchor":false},{"citing_arxiv_id":"2404.18930","citing_title":"Hallucination of Multimodal Large Language Models: A Survey","ref_index":210,"is_internal_anchor":false},{"citing_arxiv_id":"2605.00663","citing_title":"Affordance Agent Harness: Verification-Gated Skill Orchestration","ref_index":84,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/J7C24QFTMTG7HIXEQVVB52ELBS","json":"https://pith.science/pith/J7C24QFTMTG7HIXEQVVB52ELBS.json","graph_json":"https://pith.science/api/pith-number/J7C24QFTMTG7HIXEQVVB52ELBS/graph.json","events_json":"https://pith.science/api/pith-number/J7C24QFTMTG7HIXEQVVB52ELBS/events.json","paper":"https://pith.science/paper/J7C24QFT"},"agent_actions":{"view_html":"https://pith.science/pith/J7C24QFTMTG7HIXEQVVB52ELBS","download_json":"https://pith.science/pith/J7C24QFTMTG7HIXEQVVB52ELBS.json","view_paper":"https://pith.science/paper/J7C24QFT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2307.03601&json=true","fetch_graph":"https://pith.science/api/pith-number/J7C24QFTMTG7HIXEQVVB52ELBS/graph.json","fetch_events":"https://pith.science/api/pith-number/J7C24QFTMTG7HIXEQVVB52ELBS/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/J7C24QFTMTG7HIXEQVVB52ELBS/action/timestamp_anchor","attest_storage":"https://pith.science/pith/J7C24QFTMTG7HIXEQVVB52ELBS/action/storage_attestation","attest_author":"https://pith.science/pith/J7C24QFTMTG7HIXEQVVB52ELBS/action/author_attestation","sign_citation":"https://pith.science/pith/J7C24QFTMTG7HIXEQVVB52ELBS/action/citation_signature","submit_replication":"https://pith.science/pith/J7C24QFTMTG7HIXEQVVB52ELBS/action/replication_record"}},"created_at":"2026-07-05T11:20:01.418000+00:00","updated_at":"2026-07-05T11:20:01.418000+00:00"}