什么是 zero-shot learning？

模型在未经显式训练的情况下执行新任务的能力。

zero-shot learning | 概念

Q: zero-shot learning 最近有什么新动态？

traeai 已收录 1 篇与 zero-shot learning 相关的内容。最新一篇是「AI Paper Review: Language Models are Unsupervised Multitask Learners (GPT-2)」，由 freeCodeCamp.org 发布。

AI Paper Review: Language Models are Unsupervised Multitask Learners (GPT-2)

freeCodeCamp.org5月11日3193 字 (约 13 分钟)

GPT-2 demonstrated that training a large language model solely on unsupervised next-word prediction enables emergent multitask capabilities, performing translation, QA, and summarization without task-specific fine-tuning.

入选理由：GPT-2在800万网页文本上训练，参数量达15亿，首次展示零样本迁移能力。

FeaturedArticle#GPT-2#Large Language Models#Zero-Shot Learning#Transformer英文

zero-shot learning

相关材料

AI Paper Review: Language Models are Unsupervised Multitask Learners (GPT-2)

跨材料问答 · zero-shot learning