TY  - RPRT
TI  - Self-Improving is Often Sudden: Enlightenment-style Finetuning for Large-Scale Models
AU  - Jing-Xiao Liao
AU  - Tianwei Zhang
AU  - Yu-Hao Jiang
AU  - Feifei Zhang
AU  - Hang-Cheng Dong
AU  - Feng-Lei Fan
PY  - 2026
UR  - https://arxiv.org/abs/2607.13395
ID  - 2607.13395
ER  -