Eric Tao, Minhao Sun, Liyan Liu, Zhijing Wu, Xiaohan Yu, Lang Mei, Li Huang, Chao Feng · 2026 · arXiv
DOI: 10.48550/arXiv.2610.28382 · open-access PDF
Pelatihan agen pencarian multimodal membutuhkan tugas yang membuat pencarian berbasis visual diperlukan untuk menentukan jawaban. Tanda-tanda identitas teks dan bukti target yang tidak digunakan memungkinkan pendekatan dalam pertanyaan sintetis, melemahkan pengawasan untuk pencarian visual dan integrasi bukti. Kami mengusulkan RACE, kerangka kerja sintesis tugas berbasis gambar yang membangun ketergantungan dari gambar ke target dan dari bukti spesifik target ke jawaban. Abstraksi referensial (RA) menggantikan teks yang mengungkapkan identitas dengan referensi visual spesifik atau referensi relasional ambigu yang diselesaikan dengan mencari di antara
Machine translation (Meta M2M100); the English original follows.
Training multimodal search agents requires tasks that make visually grounded search necessary for determining the answer. Textual identity clues and unused target evidence allow shortcuts in synthetic questions, weakening supervision for visual search and evidence integration. We propose RACE, an image-conditioned task synthesis framework that constructs dependencies from images to targets and from target-specific evidence to answers. Referential Abstraction (RA) replaces identity-revealing text with specific visual references or ambiguous relational references resolved by searching among depi
Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese
GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.28382, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txtSources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange