{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:IQULTCBQKDNGK5KQE2VIPGCVGN","short_pith_number":"pith:IQULTCBQ","schema_version":"1.0","canonical_sha256":"4428b9883050da65755026aa879855337bb605f6fecbc66daeb3e3f6e840429a","source":{"kind":"arxiv","id":"2405.01483","version":3},"attestation_state":"computed","paper":{"title":"MANTIS: Interleaved Multi-Image Instruction Tuning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.CV","authors_text":"Cong Wei, Dongfu Jiang, Huaye Zeng, Max Ku, Qian Liu, Wenhu Chen, Xuan He","submitted_at":"2024-05-02T17:14:57Z","abstract_excerpt":"Large multimodal models (LMMs) have shown great results in single-image vision language tasks. However, their abilities to solve multi-image visual language tasks is yet to be improved. The existing LMMs like OpenFlamingo, Emu2, and Idefics gain their multi-image ability through pre-training on hundreds of millions of noisy interleaved image-text data from the web, which is neither efficient nor effective. In this paper, we aim to build strong multi-image LMMs via instruction tuning with academic-level resources. Therefore, we meticulously construct Mantis-Instruct containing 721K multi-image "},"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":"2405.01483","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2024-05-02T17:14:57Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"aa9342b4e3586aa85459d9a281653c9a2d35ce14ede5f155e46722a674507a5f","abstract_canon_sha256":"590c73692de017b640d510feff10f287a89f9c097f4ddbe141115be73041587c"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:35:48.894290Z","signature_b64":"NCmvdVDF366uIcdYDgfL9S+0X+IbVkWdOugkNvBtfNGbRSL7p2Zs4saiEqlS3glSMy5R5tNbZLjYw94eLj+/AQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4428b9883050da65755026aa879855337bb605f6fecbc66daeb3e3f6e840429a","last_reissued_at":"2026-07-05T09:35:48.893774Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:35:48.893774Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MANTIS: Interleaved Multi-Image Instruction Tuning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.CV","authors_text":"Cong Wei, Dongfu Jiang, Huaye Zeng, Max Ku, Qian Liu, Wenhu Chen, Xuan He","submitted_at":"2024-05-02T17:14:57Z","abstract_excerpt":"Large multimodal models (LMMs) have shown great results in single-image vision language tasks. However, their abilities to solve multi-image visual language tasks is yet to be improved. The existing LMMs like OpenFlamingo, Emu2, and Idefics gain their multi-image ability through pre-training on hundreds of millions of noisy interleaved image-text data from the web, which is neither efficient nor effective. In this paper, we aim to build strong multi-image LMMs via instruction tuning with academic-level resources. Therefore, we meticulously construct Mantis-Instruct containing 721K multi-image "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.01483","kind":"arxiv","version":3},"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/2405.01483/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":"2405.01483","created_at":"2026-07-05T09:35:48.893836+00:00"},{"alias_kind":"arxiv_version","alias_value":"2405.01483v3","created_at":"2026-07-05T09:35:48.893836+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.01483","created_at":"2026-07-05T09:35:48.893836+00:00"},{"alias_kind":"pith_short_12","alias_value":"IQULTCBQKDNG","created_at":"2026-07-05T09:35:48.893836+00:00"},{"alias_kind":"pith_short_16","alias_value":"IQULTCBQKDNGK5KQ","created_at":"2026-07-05T09:35:48.893836+00:00"},{"alias_kind":"pith_short_8","alias_value":"IQULTCBQ","created_at":"2026-07-05T09:35:48.893836+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":26,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.26079","citing_title":"Same Evidence, Different Answer: Auditing Order Sensitivity in Multimodal Large Language Models","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2606.17296","citing_title":"Pareto LoRA: Mitigating Modality Imbalance in Unified Multimodal Models via Pareto-Optimal Gradient Integration","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2606.13289","citing_title":"HYDRA-X: Native Unified Multimodal Models with Holistic Visual Tokenizers","ref_index":219,"is_internal_anchor":false},{"citing_arxiv_id":"2606.28551","citing_title":"DataComp-VLM: Improved Open Datasets for Vision-Language Models","ref_index":114,"is_internal_anchor":false},{"citing_arxiv_id":"2606.28551","citing_title":"DataComp-VLM: Improved Open Datasets for Vision-Language Models","ref_index":114,"is_internal_anchor":false},{"citing_arxiv_id":"2605.27959","citing_title":"ROVER: Routing Object-Centric Visual Evidence for Grounded Multi-Image Reasoning","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2605.30265","citing_title":"LoMo: Local Modality Substitution for Deeper Vision-Language Fusion","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2603.25120","citing_title":"DFLOP: A Data-driven Framework for Multimodal LLM Training Pipeline Optimization","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2605.18621","citing_title":"CrossView Suite: Harnessing Cross-view Spatial Intelligence of MLLMs with Dataset, Model and Benchmark","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2408.04840","citing_title":"mPLUG-Owl3: Towards Long Image-Sequence Understanding in Multi-Modal Large Language Models","ref_index":215,"is_internal_anchor":false},{"citing_arxiv_id":"2507.00748","citing_title":"Improving the Reasoning of Multi-Image Grounding in MLLMs via Reinforcement Learning","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2511.16518","citing_title":"MiMo-Embodied: X-Embodied Foundation Model Technical Report","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2406.09411","citing_title":"MuirBench: A Comprehensive Benchmark for Robust Multi-image Understanding","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2509.18154","citing_title":"MiniCPM-V 4.5: Cooking Efficient MLLMs via Architecture, Data, and Training Recipe","ref_index":62,"is_internal_anchor":false},{"citing_arxiv_id":"2603.04676","citing_title":"Decoding the Pulse of Reasoning VLMs in Multi-Image Understanding Tasks","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2409.17146","citing_title":"Molmo and PixMo: Open Weights and Open Data for State-of-the-Art Vision-Language Models","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2409.02813","citing_title":"MMMU-Pro: A More Robust Multi-discipline Multimodal Understanding Benchmark","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2501.13918","citing_title":"Improving Video Generation with Human Feedback","ref_index":28,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11462","citing_title":"SpatialForge: Bootstrapping 3D-Aware Spatial Reasoning from Open-World 2D Images","ref_index":52,"is_internal_anchor":false},{"citing_arxiv_id":"2604.27393","citing_title":"MiniCPM-o 4.5: Towards Real-Time Full-Duplex Omni-Modal Interaction","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2604.22498","citing_title":"CGC: Compositional Grounded Contrast for Fine-Grained Multi-Image Understanding","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2407.07895","citing_title":"LLaVA-NeXT-Interleave: Tackling Multi-image, Video, and 3D in Large Multimodal Models","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2408.03326","citing_title":"LLaVA-OneVision: Easy Visual Task Transfer","ref_index":47,"is_internal_anchor":false},{"citing_arxiv_id":"2504.10479","citing_title":"InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models","ref_index":51,"is_internal_anchor":false},{"citing_arxiv_id":"2412.05271","citing_title":"Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scaling","ref_index":100,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IQULTCBQKDNGK5KQE2VIPGCVGN","json":"https://pith.science/pith/IQULTCBQKDNGK5KQE2VIPGCVGN.json","graph_json":"https://pith.science/api/pith-number/IQULTCBQKDNGK5KQE2VIPGCVGN/graph.json","events_json":"https://pith.science/api/pith-number/IQULTCBQKDNGK5KQE2VIPGCVGN/events.json","paper":"https://pith.science/paper/IQULTCBQ"},"agent_actions":{"view_html":"https://pith.science/pith/IQULTCBQKDNGK5KQE2VIPGCVGN","download_json":"https://pith.science/pith/IQULTCBQKDNGK5KQE2VIPGCVGN.json","view_paper":"https://pith.science/paper/IQULTCBQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2405.01483&json=true","fetch_graph":"https://pith.science/api/pith-number/IQULTCBQKDNGK5KQE2VIPGCVGN/graph.json","fetch_events":"https://pith.science/api/pith-number/IQULTCBQKDNGK5KQE2VIPGCVGN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IQULTCBQKDNGK5KQE2VIPGCVGN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IQULTCBQKDNGK5KQE2VIPGCVGN/action/storage_attestation","attest_author":"https://pith.science/pith/IQULTCBQKDNGK5KQE2VIPGCVGN/action/author_attestation","sign_citation":"https://pith.science/pith/IQULTCBQKDNGK5KQE2VIPGCVGN/action/citation_signature","submit_replication":"https://pith.science/pith/IQULTCBQKDNGK5KQE2VIPGCVGN/action/replication_record"}},"created_at":"2026-07-05T09:35:48.893836+00:00","updated_at":"2026-07-05T09:35:48.893836+00:00"}