{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:2V53CWQCA26LDD42ES726NMMOI","short_pith_number":"pith:2V53CWQC","schema_version":"1.0","canonical_sha256":"d57bb15a0206bcb18f9a24bfaf358c720f3b9a2b651898105f576ef9e22b4584","source":{"kind":"arxiv","id":"2412.19326","version":2},"attestation_state":"computed","paper":{"title":"Task Preference Optimization: Improving Multimodal Large Language Models with Vision Task Alignment","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Chenting Wang, Kunchang Li, Limin Wang, Xiangyu Zeng, Xinhao Li, Yali Wang, Yinan He, Yi Wang, Yu Qiao, Zhilin Li, Ziang Yan, Zilei Wang","submitted_at":"2024-12-26T18:56:05Z","abstract_excerpt":"Current multimodal large language models (MLLMs) struggle with fine-grained or precise understanding of visuals although they give comprehensive perception and reasoning in a spectrum of vision applications. Recent studies either develop tool-using or unify specific visual tasks into the autoregressive framework, often at the expense of overall multimodal performance. To address this issue and enhance MLLMs with visual tasks in a scalable fashion, we propose Task Preference Optimization (TPO), a novel method that utilizes differentiable task preferences derived from typical fine-grained visual"},"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":"2412.19326","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CV","submitted_at":"2024-12-26T18:56:05Z","cross_cats_sorted":[],"title_canon_sha256":"2302667309091af77e42215b883d28d65ad3d4f6e56cc4b789df9c45f28f7cef","abstract_canon_sha256":"9c8fc5c449db424400b21af215898f95a371244a931f65a8ae6676ea77288123"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:29:04.952068Z","signature_b64":"jV4uUokz2hFxTf0j3T5eSPnsFb/VzSGhKGW0kv//GWFg8KbjKA+mxu3/qhHy5h4TAQxNggFS+fGh9bgjd/tNDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d57bb15a0206bcb18f9a24bfaf358c720f3b9a2b651898105f576ef9e22b4584","last_reissued_at":"2026-07-05T11:29:04.951561Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:29:04.951561Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Task Preference Optimization: Improving Multimodal Large Language Models with Vision Task Alignment","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Chenting Wang, Kunchang Li, Limin Wang, Xiangyu Zeng, Xinhao Li, Yali Wang, Yinan He, Yi Wang, Yu Qiao, Zhilin Li, Ziang Yan, Zilei Wang","submitted_at":"2024-12-26T18:56:05Z","abstract_excerpt":"Current multimodal large language models (MLLMs) struggle with fine-grained or precise understanding of visuals although they give comprehensive perception and reasoning in a spectrum of vision applications. Recent studies either develop tool-using or unify specific visual tasks into the autoregressive framework, often at the expense of overall multimodal performance. To address this issue and enhance MLLMs with visual tasks in a scalable fashion, we propose Task Preference Optimization (TPO), a novel method that utilizes differentiable task preferences derived from typical fine-grained visual"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2412.19326","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/2412.19326/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":"2412.19326","created_at":"2026-07-05T11:29:04.951623+00:00"},{"alias_kind":"arxiv_version","alias_value":"2412.19326v2","created_at":"2026-07-05T11:29:04.951623+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2412.19326","created_at":"2026-07-05T11:29:04.951623+00:00"},{"alias_kind":"pith_short_12","alias_value":"2V53CWQCA26L","created_at":"2026-07-05T11:29:04.951623+00:00"},{"alias_kind":"pith_short_16","alias_value":"2V53CWQCA26LDD42","created_at":"2026-07-05T11:29:04.951623+00:00"},{"alias_kind":"pith_short_8","alias_value":"2V53CWQC","created_at":"2026-07-05T11:29:04.951623+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2501.12386","citing_title":"InternVideo2.5: Empowering Video MLLMs with Long and Rich Context Modeling","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2504.06958","citing_title":"VideoChat-R1: Enhancing Spatio-Temporal Perception via Reinforcement Fine-Tuning","ref_index":33,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2V53CWQCA26LDD42ES726NMMOI","json":"https://pith.science/pith/2V53CWQCA26LDD42ES726NMMOI.json","graph_json":"https://pith.science/api/pith-number/2V53CWQCA26LDD42ES726NMMOI/graph.json","events_json":"https://pith.science/api/pith-number/2V53CWQCA26LDD42ES726NMMOI/events.json","paper":"https://pith.science/paper/2V53CWQC"},"agent_actions":{"view_html":"https://pith.science/pith/2V53CWQCA26LDD42ES726NMMOI","download_json":"https://pith.science/pith/2V53CWQCA26LDD42ES726NMMOI.json","view_paper":"https://pith.science/paper/2V53CWQC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2412.19326&json=true","fetch_graph":"https://pith.science/api/pith-number/2V53CWQCA26LDD42ES726NMMOI/graph.json","fetch_events":"https://pith.science/api/pith-number/2V53CWQCA26LDD42ES726NMMOI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2V53CWQCA26LDD42ES726NMMOI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2V53CWQCA26LDD42ES726NMMOI/action/storage_attestation","attest_author":"https://pith.science/pith/2V53CWQCA26LDD42ES726NMMOI/action/author_attestation","sign_citation":"https://pith.science/pith/2V53CWQCA26LDD42ES726NMMOI/action/citation_signature","submit_replication":"https://pith.science/pith/2V53CWQCA26LDD42ES726NMMOI/action/replication_record"}},"created_at":"2026-07-05T11:29:04.951623+00:00","updated_at":"2026-07-05T11:29:04.951623+00:00"}