1 paper · 1 filter
Shahriar Golchin, Mihai Surdeanu, Steven Bethard +2
In-context learning (ICL) has proven to be an effective strategy for improving the performance of large language models (LLMs) with no additional training. However, the exact mecha…