[CL] You Only Fine-tune Once：Many-Shot In-Context Fine-Tuning for Large Language Model
[Google & University of Florida]
https://arxiv.org/abs/2506.11103