按年被提到
20212022 · 1 次2026
大模型与语言模型›大模型预训练与缩放·arXiv:2109.01652 · cs.CL

Finetuned Language Models Are Zero-Shot Learners

Jason Wei, Maarten Bosma, Vincent Y. Zhao, Kelvin Guu, Adams Wei Yu, Brian Lester, Nan Du, Andrew M. Dai, Quoc V. Le · 2021
开始阅读 ↗PDFalphaXiv 讨论
被提及 · 2 人首次 2022-05-02 · 最近 2023-07-11
2不同的人
2活跃月份
2讨论串
1,909被引(OpenAlex)
arXiv 摘要(候选层不写导读)

This paper explores a simple method for improving the zero-shot learning abilities of language models. We show that instruction tuning -- finetuning language models on a collection of tasks described via instructions -- substantially improves zero-shot performance on unseen tasks. We take a 137B parameter pretrained language model and instruction-tune it on over 60 NLP tasks verbalized via natural language instruction templates. We evaluate this instruction-tuned model, which we call FLAN, on unseen task types. FLAN substantially improves the performance of its unmodified counterpart and surpasses zero-shot 175B GPT-3 on 20 of 25 tasks that we evaluate. FLAN even outperforms few-shot GPT-3 by a large margin on ANLI, RTE, BoolQ, AI2-ARC, OpenbookQA, and StoryCloze. Ablation studies reveal that number of finetuning datasets, model scale, and natural language instructions are key to the success of instruction tuning.

候选层:至少 2 个不同的人、在 2 个不同月份提到过,还没到主书架的 3 人 3 个月。没有做立场标注,所以只写“被提及”。

HN 书评

推荐在前,同类按时间 · 每篇最多保留 20 条,统计用全部提及
评论joshuamorton · 2022-05

“The controversy here is specifically around Stochastic Parrots, by among others Timnit Gebru and "Schmargaret Smitchell". Timnit had recently been fired by Google for a controversy that started when leadership prevented her and Margaret from being coauthors on the paper (which is critical of Google in some ways) and Mitchell was fired a few months later. …”

Google fires another AI researcher who reportedly challenged findings原评论 ↗
评论radarsat1 · 2023-07

“It comes from this paper [0], and I believe the idea is that the LLM was not trained on the task in question, but is able to do it with only instructions (zero shot) or with one or a few examples (few shot). The paper rightfully points out the unexpected fact that the model is only trained to predict the next word and yet can follow arbitrary instructions…”

What we learned from using GPT for 500k+ classifications原评论 ↗