Long-Horizon Scaling: How Model Capabilities Shape the Returns to Computation
It is found that starting performance and subsequent growth are associated with different capabilities: within a task category, similar early scores can precede different later gains, and a continuation policy to save time and compute with limited score loss is derived.
Hao-Yu Zheng, Zheng-Yu Chen, Huai-Sheng Zhu et al.
· 0 citations