[Adapt] [Seminar] Introduction to In-Context Learning

[任思宇]

Hi Adapters,

Large language models(LLMs) have taken the NLP world by storm in recent years. One intriguing emergent ability of LLMs is called in-context learning(ICL), where the model learns to make prediction for a test input by conditioning on a handfulof task-specific examples without any parameter updates. Such a novel paradigm has attained much attention due to its efficiency when handling multiple downstream tasks. In this talk, I will present an introduction about ICL, including the background, advantages compared to vanilla fine-tuning, ways to improve ICL performance, and what makes ICL work.

Hope you find this talk interesting and useful.

Time: Wed 4:00 pm
Venue: SEIEE 3-404

Best Regards,
Roy
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://cs.sjtu.edu.cn/pipermail/adapt/attachments/20230222/c7c8cfab/attachment.htm>