[Adapt] An Introduction to Instruction Tuning

[王宇飞]

Hi, Adapters

This week I will give a talk about Instruction Tuning, which is an emergent paradigm in NLP wherein natural language instructions are leveraged with language models to induce zero-shot performances on unseen tasks. Unlike prompt, which aims to trigger the completion ability of the model, instruction tuning aims to trigger the understanding ability of the model by giving more clear instructions to help the model to give right feedback. 
In this talk, I will introduce several papers, explaining the basic idea of Instructing Tuning, key influencing factors and how to construct new Instructing datasets from plain texts. 

Hope you will enjoy it,
Arthur