ARC Prize：ARC-AGI 三项基准分数强相关
Plotting ARC-AGI model scores reveals a strong relationship across all three benchmarks.
AI 导读
ARC Prize 绘制 ARC-AGI 模型得分后发现，三项基准之间存在强关联：幂曲线 v2 = v1^4.53 可解释约 98% 的 ARC-AGI-2 得分变化，在 ARC-AGI-1 上持平的模型，在 ARC-AGI-2 上仍可能相差超过 20 个百分点。ARC-AGI-3 上也有类似趋势，但 Gemini 4 Argon 低推理档是明显例外：ARC-AGI-2 仅 42.1%，ARC-AGI-3 却达 97.5%。
正文
Plotting ARC-AGI model scores reveals a strong relationship across all three benchmarks.
For example, the power curve v2 = v1^4.53 explains about 98% of the variation in ARC-AGI-2 scores. Models tied on ARC-AGI-1 can still differ by over 20 percentage points on ARC-AGI-2. https://t.co/7uG7hNXAwo