揭秘夸克首个高考志愿大模型！蒸馏数百名人类专家经验、Agent 可完整生成志愿报告

Core Viewpoint - Quark has launched the first high school entrance examination (Gaokao) volunteer filling model in China, providing personalized decision-making services for students during the college application process [1][3]. Group 1: Features of the Quark Gaokao Volunteer Model - The model operates with expert-level decision-making capabilities, offering tailored volunteer filling services based on students' scores, interests, family background, and regional preferences [3][4]. - It utilizes a task planning-execution-check-reflection reasoning process to generate comprehensive reports that include strategies for application, recommended schools, and majors [3][4]. - The "Deep Search" function allows users to input complex queries, which the model breaks down into specific needs, ensuring targeted and in-depth responses [4][11]. Group 2: Training and Data Sources - The model is built on a multi-stage, high-complexity training paradigm, integrating self-supervised semantic modeling and expert-guided strategy refinement [7][9]. - It has structured the communication and decision-making processes of experienced volunteer planners, converting thousands of real expert reasoning chains into high-quality supervised data for deep learning [9][11]. - The knowledge base of the model is the largest in China, covering over 2,900 universities and nearly 1,600 undergraduate programs, ensuring comprehensive and authoritative data for decision-making [11][10]. Group 3: Optimization and Feedback Mechanism - The model employs a closed-loop optimization mechanism that incorporates simulated application scenarios, expert feedback, and strategy scoring to continuously refine its outputs [9][11]. - It aims to provide a comprehensive reference for every student and family by leveraging its advantages in information processing and understanding user needs [11].