Out-of-domain generalization from rephrasability and stability
Produced by Generalization Bounds for Autoregressive Processes and In-Context Learning
In-context learning
The adaptation of a model’s predictions to examples or instructions supplied in its input context without updating model parameters.
The adaptation of a model's predictions to examples or instructions supplied in its input context without updating model parameters.
It is studied both as observable behavior and as an inference procedure implemented by a sequence model.