In-Context Learning through the Bayesian Prism

arXiv:2306.04891v2 Announce Type: replace
Abstract: In-context learning (ICL) is one of the surprising and useful features of large language models and subject of intense research. Recently, stylized meta-learning-like ICL setups have been devised that train transformers on sequences of input-output pairs $(x, f(x))$. The function $f$ comes from a function class and generalization is checked by evaluating on sequences generated from unseen functions from the same class. One of the main discoveries in this line of research has been that …

abstract arxiv bayesian class context cs.cl cs.lg features function in-context learning input-output language language models large language large language models meta meta-learning research through train transformers type