{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:5HJYODELBZFGMT5Y7JUBIMQIKI","short_pith_number":"pith:5HJYODEL","schema_version":"1.0","canonical_sha256":"e9d3870c8b0e4a664fb8fa68143208521f98ab020bb761213fd2cd30efb4de5b","source":{"kind":"arxiv","id":"2503.09445","version":2},"attestation_state":"computed","paper":{"title":"Astrea: A MOE-based Visual Understanding Model with Progressive Alignment","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Cong Lin, Hao Li, Hongshun Qiu, Jiaqi Duan, Jiayang Xu, Junyu Lu, Shengpeng Ji, Sihang Cai, Sijing Li, Songxin Zhang, Xiaoda Yang, Xudong Tang, Zejian Xie, Zhuoyang Song, Ziyue Jiang","submitted_at":"2025-03-12T14:44:52Z","abstract_excerpt":"Vision-Language Models (VLMs) based on Mixture-of-Experts (MoE) architectures have emerged as a pivotal paradigm in multimodal understanding, offering a powerful framework for integrating visual and linguistic information. However, the increasing complexity and diversity of tasks present significant challenges in coordinating load balancing across heterogeneous visual experts, where optimizing one specialist's performance often compromises others' capabilities. To address task heterogeneity and expert load imbalance, we propose Astrea, a novel multi-expert collaborative VLM architecture based "},"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":"2503.09445","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CV","submitted_at":"2025-03-12T14:44:52Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"43b8198827872b6a0b19c7751456ac664a4b0283167cd55e9cc9ea54532e87bf","abstract_canon_sha256":"ce21ef96c66b35cf506a47f9f252bd8befa118bc1877d0722b016359fef4c427"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:42:15.699059Z","signature_b64":"QEdi9sBRRRuL1o7VMYHjwYm/tgKE+nsgvda+V11n3OLLEl+J7AhWfSkatPojA9CiS2PHpQvMyjqNi/wSHmhLBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e9d3870c8b0e4a664fb8fa68143208521f98ab020bb761213fd2cd30efb4de5b","last_reissued_at":"2026-07-05T10:42:15.698556Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:42:15.698556Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Astrea: A MOE-based Visual Understanding Model with Progressive Alignment","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Cong Lin, Hao Li, Hongshun Qiu, Jiaqi Duan, Jiayang Xu, Junyu Lu, Shengpeng Ji, Sihang Cai, Sijing Li, Songxin Zhang, Xiaoda Yang, Xudong Tang, Zejian Xie, Zhuoyang Song, Ziyue Jiang","submitted_at":"2025-03-12T14:44:52Z","abstract_excerpt":"Vision-Language Models (VLMs) based on Mixture-of-Experts (MoE) architectures have emerged as a pivotal paradigm in multimodal understanding, offering a powerful framework for integrating visual and linguistic information. However, the increasing complexity and diversity of tasks present significant challenges in coordinating load balancing across heterogeneous visual experts, where optimizing one specialist's performance often compromises others' capabilities. To address task heterogeneity and expert load imbalance, we propose Astrea, a novel multi-expert collaborative VLM architecture based "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.09445","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/2503.09445/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":"2503.09445","created_at":"2026-07-05T10:42:15.698615+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.09445v2","created_at":"2026-07-05T10:42:15.698615+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.09445","created_at":"2026-07-05T10:42:15.698615+00:00"},{"alias_kind":"pith_short_12","alias_value":"5HJYODELBZFG","created_at":"2026-07-05T10:42:15.698615+00:00"},{"alias_kind":"pith_short_16","alias_value":"5HJYODELBZFGMT5Y","created_at":"2026-07-05T10:42:15.698615+00:00"},{"alias_kind":"pith_short_8","alias_value":"5HJYODEL","created_at":"2026-07-05T10:42:15.698615+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.31201","citing_title":"ExPLoRe: Expert Patch-Level Loss Routing for Multi-Objective Masked Image Modeling","ref_index":34,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5HJYODELBZFGMT5Y7JUBIMQIKI","json":"https://pith.science/pith/5HJYODELBZFGMT5Y7JUBIMQIKI.json","graph_json":"https://pith.science/api/pith-number/5HJYODELBZFGMT5Y7JUBIMQIKI/graph.json","events_json":"https://pith.science/api/pith-number/5HJYODELBZFGMT5Y7JUBIMQIKI/events.json","paper":"https://pith.science/paper/5HJYODEL"},"agent_actions":{"view_html":"https://pith.science/pith/5HJYODELBZFGMT5Y7JUBIMQIKI","download_json":"https://pith.science/pith/5HJYODELBZFGMT5Y7JUBIMQIKI.json","view_paper":"https://pith.science/paper/5HJYODEL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.09445&json=true","fetch_graph":"https://pith.science/api/pith-number/5HJYODELBZFGMT5Y7JUBIMQIKI/graph.json","fetch_events":"https://pith.science/api/pith-number/5HJYODELBZFGMT5Y7JUBIMQIKI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5HJYODELBZFGMT5Y7JUBIMQIKI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5HJYODELBZFGMT5Y7JUBIMQIKI/action/storage_attestation","attest_author":"https://pith.science/pith/5HJYODELBZFGMT5Y7JUBIMQIKI/action/author_attestation","sign_citation":"https://pith.science/pith/5HJYODELBZFGMT5Y7JUBIMQIKI/action/citation_signature","submit_replication":"https://pith.science/pith/5HJYODELBZFGMT5Y7JUBIMQIKI/action/replication_record"}},"created_at":"2026-07-05T10:42:15.698615+00:00","updated_at":"2026-07-05T10:42:15.698615+00:00"}