sakamoro/agi-olympics-v3

AGI Olympics V3是一个用于评估人工通用智能（AGI）能力的综合基准测试，涵盖四个层级：1. 自我意识与自我改进（4项测试）；2. 核心能力（4项测试）；3. 意识（1项测试）；4. 长期记忆（4项测试）。数据集包含测试问题、评估协议和样本数据，支持英语和日语。采用黑盒评估方法，基于可观察行为和输出进行评估，确保科学有效性和可重复性。

AGI Olympics V3 is a comprehensive benchmark for evaluating Artificial General Intelligence (AGI) capabilities across four tiers: Tier 1: Self-Awareness & Self-Improvement (4 tests), Tier 2: Core Capabilities (4 tests), Tier 3: Consciousness (1 test), and Tier 4: Long-Term Memory (4 tests). The dataset contains test questions, evaluation protocols, and sample data, with full support for English and Japanese. It employs a black-box evaluation methodology, relying solely on observable behavior and outputs for assessment, ensuring scientific validity and reproducibility.