{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:5QUREKRAZJMEP566VQEXDP2NHE","short_pith_number":"pith:5QUREKRA","schema_version":"1.0","canonical_sha256":"ec29122a20ca5847f7deac0971bf4d393e7bb268c9e09d3360d020d0c708b9b6","source":{"kind":"arxiv","id":"2407.00634","version":2},"attestation_state":"computed","paper":{"title":"Tarsier: Recipes for Training and Evaluating Large Video Description Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Haomiao Sun, Jiawei Wang, Liping Yuan, Yuchen Zhang","submitted_at":"2024-06-30T09:21:01Z","abstract_excerpt":"Generating fine-grained video descriptions is a fundamental challenge in video understanding. In this work, we introduce Tarsier, a family of large-scale video-language models designed to generate high-quality video descriptions. Tarsier employs CLIP-ViT to encode frames separately and then uses an LLM to model temporal relationships. Despite its simple architecture, we demonstrate that with a meticulously designed two-stage training procedure, the Tarsier models exhibit substantially stronger video description capabilities than any existing open-source model, showing a $+51.4\\%$ advantage in "},"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":"2407.00634","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2024-06-30T09:21:01Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"1dbbb9dd25ebd3a6d3d7a676c14ded3737cabfcb6bc8b8aabf29d5026efe7d18","abstract_canon_sha256":"b6c7ad42505f9d216e01b6ccaa4ae1b04c479ec88979fd9eded34594cda83a61"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:10:48.635389Z","signature_b64":"7VSrVGMHhTBzV61nWheiMVMxolL6+Qh/AwkA5HiNunTOh6XtTyyyLjLFCyMPlVF33ptqmog4aSUnQWBhTHNaAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ec29122a20ca5847f7deac0971bf4d393e7bb268c9e09d3360d020d0c708b9b6","last_reissued_at":"2026-07-05T09:10:48.634940Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:10:48.634940Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Tarsier: Recipes for Training and Evaluating Large Video Description Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Haomiao Sun, Jiawei Wang, Liping Yuan, Yuchen Zhang","submitted_at":"2024-06-30T09:21:01Z","abstract_excerpt":"Generating fine-grained video descriptions is a fundamental challenge in video understanding. In this work, we introduce Tarsier, a family of large-scale video-language models designed to generate high-quality video descriptions. Tarsier employs CLIP-ViT to encode frames separately and then uses an LLM to model temporal relationships. Despite its simple architecture, we demonstrate that with a meticulously designed two-stage training procedure, the Tarsier models exhibit substantially stronger video description capabilities than any existing open-source model, showing a $+51.4\\%$ advantage in "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.00634","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/2407.00634/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":"2407.00634","created_at":"2026-07-05T09:10:48.634992+00:00"},{"alias_kind":"arxiv_version","alias_value":"2407.00634v2","created_at":"2026-07-05T09:10:48.634992+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.00634","created_at":"2026-07-05T09:10:48.634992+00:00"},{"alias_kind":"pith_short_12","alias_value":"5QUREKRAZJME","created_at":"2026-07-05T09:10:48.634992+00:00"},{"alias_kind":"pith_short_16","alias_value":"5QUREKRAZJMEP566","created_at":"2026-07-05T09:10:48.634992+00:00"},{"alias_kind":"pith_short_8","alias_value":"5QUREKRA","created_at":"2026-07-05T09:10:48.634992+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":25,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.24636","citing_title":"CineCap: Structured Reasoning with Spatio-Temporal Anchors for Cinematographic Video Captioning","ref_index":48,"is_internal_anchor":false},{"citing_arxiv_id":"2606.26196","citing_title":"From Structure to Synergy: A Survey of Vision-Language Perception Paradigm Evolution in Multimodal Large Language Models","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2607.01667","citing_title":"Temporal and Cross-Modal Alignment for Enhanced Audiovisual Video Captioning","ref_index":56,"is_internal_anchor":false},{"citing_arxiv_id":"2606.12195","citing_title":"InternVideo3: Agentify Foundation Models with Multimodal Contextual Reasoning","ref_index":297,"is_internal_anchor":false},{"citing_arxiv_id":"2606.09181","citing_title":"Counterfactual Reasoning for Fine-Grained Evidence Disentanglement in VideoQA","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2606.09393","citing_title":"CapRL++: Unified Reinforcement Learning with Verifiable Rewards for Dense Image and Video Captioning","ref_index":76,"is_internal_anchor":false},{"citing_arxiv_id":"2606.07433","citing_title":"Watch, Remember, Reason: Human-View Video Understanding with MLLMs","ref_index":92,"is_internal_anchor":false},{"citing_arxiv_id":"2606.06853","citing_title":"MotionEnhancer: Leveraging Video Diffusion for Motion-Enhanced Vision-Language Models","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2606.05748","citing_title":"UNIVID: Unified Vision-Language Model for Video Moderation","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2606.07643","citing_title":"AVI-Bench: Toward Human-like Audio-Visual Intelligence of Omni-MLLMs","ref_index":80,"is_internal_anchor":false},{"citing_arxiv_id":"2606.29531","citing_title":"MotionAtlas: Detailed Region Captioning for Motion-Centric Videos","ref_index":34,"is_internal_anchor":false},{"citing_arxiv_id":"2605.28023","citing_title":"VCap: Hypergeometric Rewards for Weak-to-Strong Visual Captioning","ref_index":46,"is_internal_anchor":false},{"citing_arxiv_id":"2412.00131","citing_title":"Open-Sora Plan: Open-Source Large Video Generation Model","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2605.20165","citing_title":"CaMo: Camera Motion Grounded Evaluation and Training for Vision-Language Models","ref_index":46,"is_internal_anchor":false},{"citing_arxiv_id":"2505.20715","citing_title":"MUSEG: Reinforcing Video Temporal Understanding via Timestamp-Aware Multi-Segment Grounding","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2512.13511","citing_title":"Adapting MLLMs for Nuanced Video Retrieval","ref_index":69,"is_internal_anchor":false},{"citing_arxiv_id":"2602.00181","citing_title":"CamReasoner: Reinforcing Camera Movement Understanding via Structured Spatial Reasoning","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2504.06958","citing_title":"VideoChat-R1: Enhancing Spatio-Temporal Perception via Reinforcement Fine-Tuning","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2603.03944","citing_title":"SCP: Spatial Causal Prediction in Video","ref_index":44,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14569","citing_title":"Bridging Brain and Semantics: A Hierarchical Framework for Semantically Enhanced fMRI-to-Video Reconstruction","ref_index":100,"is_internal_anchor":false},{"citing_arxiv_id":"2604.02891","citing_title":"Progressive Video Condensation with MLLM Agent for Long-form Video Understanding","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2604.21718","citing_title":"Building a Precise Video Language with Human-AI Oversight","ref_index":61,"is_internal_anchor":false},{"citing_arxiv_id":"2505.07062","citing_title":"Seed1.5-VL Technical Report","ref_index":140,"is_internal_anchor":false},{"citing_arxiv_id":"2406.07476","citing_title":"VideoLLaMA 2: Advancing Spatial-Temporal Modeling and Audio Understanding in Video-LLMs","ref_index":48,"is_internal_anchor":false},{"citing_arxiv_id":"2410.02713","citing_title":"LLaVA-Video: Video Instruction Tuning With Synthetic Data","ref_index":48,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5QUREKRAZJMEP566VQEXDP2NHE","json":"https://pith.science/pith/5QUREKRAZJMEP566VQEXDP2NHE.json","graph_json":"https://pith.science/api/pith-number/5QUREKRAZJMEP566VQEXDP2NHE/graph.json","events_json":"https://pith.science/api/pith-number/5QUREKRAZJMEP566VQEXDP2NHE/events.json","paper":"https://pith.science/paper/5QUREKRA"},"agent_actions":{"view_html":"https://pith.science/pith/5QUREKRAZJMEP566VQEXDP2NHE","download_json":"https://pith.science/pith/5QUREKRAZJMEP566VQEXDP2NHE.json","view_paper":"https://pith.science/paper/5QUREKRA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2407.00634&json=true","fetch_graph":"https://pith.science/api/pith-number/5QUREKRAZJMEP566VQEXDP2NHE/graph.json","fetch_events":"https://pith.science/api/pith-number/5QUREKRAZJMEP566VQEXDP2NHE/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5QUREKRAZJMEP566VQEXDP2NHE/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5QUREKRAZJMEP566VQEXDP2NHE/action/storage_attestation","attest_author":"https://pith.science/pith/5QUREKRAZJMEP566VQEXDP2NHE/action/author_attestation","sign_citation":"https://pith.science/pith/5QUREKRAZJMEP566VQEXDP2NHE/action/citation_signature","submit_replication":"https://pith.science/pith/5QUREKRAZJMEP566VQEXDP2NHE/action/replication_record"}},"created_at":"2026-07-05T09:10:48.634992+00:00","updated_at":"2026-07-05T09:10:48.634992+00:00"}