按年被提到
20232023 · 2 次2026
大模型与语言模型›大模型推理与思维链·arXiv:2304.03262 · cs.AI

When do you need Chain-of-Thought Prompting for ChatGPT?

Jiuhai Chen, Lichang Chen, Heng Huang, Tianyi Zhou · 2023
开始阅读 ↗PDFalphaXiv 讨论
被提及 · 2 人首次 2023-04-09 · 最近 2023-05-16
2不同的人
2活跃月份
2讨论串
46被引(OpenAlex)
arXiv 摘要(候选层不写导读)

Chain-of-Thought (CoT) prompting can effectively elicit complex multi-step reasoning from Large Language Models~(LLMs). For example, by simply adding CoT instruction ``Let's think step-by-step'' to each input query of MultiArith dataset, GPT-3's accuracy can be improved from 17.7\% to 78.7\%. However, it is not clear whether CoT is still effective on more recent instruction finetuned (IFT) LLMs such as ChatGPT. Surprisingly, on ChatGPT, CoT is no longer effective for certain tasks such as arithmetic reasoning while still keeping effective on other reasoning tasks. Moreover, on the former tasks, ChatGPT usually achieves the best performance and can generate CoT even without being instructed to do so. Hence, it is plausible that ChatGPT has already been trained on these tasks with CoT and thus memorized the instruction so it implicitly follows such an instruction when applied to the same queries, even without CoT. Our analysis reflects a potential risk of overfitting/bias toward instructions introduced in IFT, which becomes more common in training LLMs. In addition, it indicates possible leakage of the pretraining recipe, e.g., one can verify whether a dataset and instruction were used in training ChatGPT. Our experiments report new baseline results of ChatGPT on a variety of reasoning tasks and shed novel insights into LLM's profiling, instruction memorization, and pretraining dataset leakage.

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thewataccount · 2023-05

“Woah I think you're right. It's output looks suspiciously like Chain of thought too https://arxiv.org/abs/2304.03262 Here's a question - do we consider the intelligence of the model itself, or do we consider the possible intelligence of the entire system including agents? Because yeah if it can decide when it should chain stuff together then it's getti…”

Ask HN: Is this a good test of Chat AI reasoning ability?原评论 ↗
被提交为 story · 1 次