固有のタスクに応じた訓練やファインチューニングが永続的なものであるのに対し、コンテキスト内学習による学習は一時的である。一時的なコンテキストやバイアスは、(事前)訓練データセットにすでに存在するものを除き、会話から会話へと持ち越されることはない[18]。トランスフォーマー層内における「メサ最適化(mesa-optimization)」[19][20]のこの結果は、メタ学習または「学習する学習(learning to learn)」の一形態である[21]。
「テキストから画像」モデルの中には、特定のアーティストの作風を名前から模倣できるものがある。たとえば、Stable DiffusionやMidjourneyのプロンプトでは、「in the style of Greg Rutkowski」という語句が、ポーランドのデジタルアーティストGreg Rutkowskiの特徴的な作風の画像を生成する[55]。
ネガティブプロンプト
「テキストから画像」モデルは「否定」を文字どおりには理解しない。たとえば、「a party with no cake」(ケーキのないパーティー)というプロンプトは、ケーキを含む画像を生成する可能性がある[56]。代替策として否定プロンプトを使用することで、結果の画像に表示されるべきでない用語を別のプロンプトで指定することができる[57]。一般的なアプローチは、画像の否定プロンプトに、ugly, boring, bad anatomy(醜い、退屈、悪い構成)などの一般的な望ましくない用語を含めることである。
^ abc“Stable Diffusion Prompt Book” (2022年10月28日). 2023年8月7日閲覧。 “"Prompt engineering is the process of structuring words that can be interpreted and understood by a text-to-image model. Think of it as the language you need to speak in order to tell an AI model what to draw."”
^Wei, Jason; Tay, Yi; Bommasani, Rishi; Raffel, Colin; Zoph, Barret; Borgeaud, Sebastian; Yogatama, Dani; Bosma, Maarten; Zhou, Denny; Metzler, Donald; Chi, Ed H.; Hashimoto, Tatsunori; Vinyals, Oriol; Liang, Percy; Dean, Jeff; Fedus, William (31 August 2022). "Emergent Abilities of Large Language Models". arXiv:2206.07682 [cs.CL]. In prompting, a pre-trained language model is given a prompt (e.g. a natural language instruction) of a task and completes the response without any further training or gradient updates to its parameters... The ability to perform a task via few-shot prompting is emergent when a model has random performance until a certain scale, after which performance increases to well-above random
^“Language Models are Unsupervised Multitask Learners”. OpenAI blog (2019年). 2023年8月20日閲覧。 “"We demonstrate language models can perform down-stream tasks in a zero-shot setting - without any parameter or architecture modification"”
^OpenAI (2022年11月30日). “Introducing ChatGPT”. OpenAI Blog. 2023年8月16日閲覧。 “"what is the fermat's little theorem"”
^ abRobinson, Reid (August 3, 2023). “How to write an effective GPT-3 or GPT-4 prompt”. Zapier. 2023年8月14日閲覧。 “"Basic prompt: 'Write a poem about leaves falling.' Better prompt: 'Write a poem in the style of Edgar Allan Poe about leaves falling.'”
^Garg, Shivam; Tsipras, Dimitris; Liang, Percy; Valiant, Gregory (2022). "What Can Transformers Learn In-Context? A Case Study of Simple Function Classes". arXiv:2208.01066 [cs.CL]。
^Brown, Tom; Mann, Benjamin; Ryder, Nick; Subbiah, Melanie; Kaplan, Jared D.; Dhariwal, Prafulla; Neelakantan, Arvind (2020). “Language models are few-shot learners”. Advances in neural information processing systems33: 1877-1901.
^Wiggers, Kyle (2023年6月12日). “Meta open sources an AI-powered music generator”. TechCrunch. 2023年8月15日閲覧。 “Next, I gave a more complicated prompt to attempt to throw MusicGen for a loop: "Lo-fi slow BPM electro chill with organic samples."”
^Caballero, Ethan; Gupta, Kshitij; Rish, Irina; Krueger, David (2022). "Broken Neural Scaling Laws". International Conference on Learning Representations (ICLR), 2023.
^Wei, Jason; Tay, Yi; Bommasani, Rishi; Raffel, Colin; Zoph, Barret; Borgeaud, Sebastian; Yogatama, Dani; Bosma, Maarten; Zhou, Denny; Metzler, Donald; Chi, Ed H.; Hashimoto, Tatsunori; Vinyals, Oriol; Liang, Percy; Dean, Jeff; Fedus, William (31 August 2022). "Emergent Abilities of Large Language Models". arXiv:2206.07682 [cs.CL]。
^ abcdefWei, Jason; Wang, Xuezhi; Schuurmans, Dale; Bosma, Maarten; Ichter, Brian; Xia, Fei; Chi, Ed H.; Le, Quoc V.; Zhou, Denny (31 October 2022). "Chain-of-Thought Prompting Elicits Reasoning in Large Language Models" (英語). arXiv:2201.11903 [cs.CL]。
^“How AI Knows Things No One Told It”. Scientific American. 17 May 2023閲覧。 “"By the time you type a query into ChatGPT, the network should be fixed; unlike humans, it should not continue to learn. So it came as a surprise that LLMs do, in fact, learn from their users' prompts—an ability known as in-context learning."”
^Johannes von Oswald; Niklasson, Eyvind; Randazzo, Ettore; Sacramento, Joテ」o; Mordvintsev, Alexander; Zhmoginov, Andrey; Vladymyrov, Max (2022). "Transformers learn in-context by gradient descent". arXiv:2212.07677 [cs.LG]. Thus we show how trained Transformers become mesa-optimizers i.e. learn models by gradient descent in their forward pass
^“Mesa-Optimization”. 17 May 2023閲覧。 “"Mesa-Optimization is the situation that occurs when a learned model (such as a neural network) is itself an optimizer."”
^Garg, Shivam; Tsipras, Dimitris; Liang, Percy; Valiant, Gregory (2022). "What Can Transformers Learn In-Context? A Case Study of Simple Function Classes". arXiv:2208.01066 [cs.CL]. Training a model to perform in-context learning can be viewed as an instance of the more general learning-to-learn or meta-learning paradigm
^Sanh, Victor; et al. (2021). "Multitask Prompted Training Enables Zero-Shot Task Generalization". arXiv:2110.08207 [cs.LG]。
^Bach, Stephen H.; Sanh, Victor; Yong, Zheng-Xin; Webson, Albert; Raffel, Colin; Nayak, Nihal V.; Sharma, Abheesht; Kim, Taewoon; M Saiful Bari; Fevry, Thibault; Alyafeai, Zaid; Dey, Manan; Santilli, Andrea; Sun, Zhiqing; Ben-David, Srulik; Xu, Canwen; Chhablani, Gunjan; Wang, Han; Jason Alan Fries; Al-shaibani, Maged S.; Sharma, Shanya; Thakker, Urmish; Almubarak, Khalid; Tang, Xiangru; Radev, Dragomir; Mike Tian-Jian Jiang; Rush, Alexander M. (2022). "PromptSource: An Integrated Development Environment and Repository for Natural Language Prompts". arXiv:2202.01279 [cs.LG]。
^Wang, Xuezhi; Wei, Jason; Schuurmans, Dale; Le, Quoc; Chi, Ed; Narang, Sharan; Chowdhery, Aakanksha; Zhou, Denny (1 March 2022). "Self-Consistency Improves Chain of Thought Reasoning in Language Models". arXiv:2203.11171 [cs.CL]。
^Diao, Shizhe; Wang, Pengcheng; Lin, Yong; Zhang, Tong (1 February 2023). "Active Prompting with Chain-of-Thought for Large Language Models". arXiv:2302.12246 [cs.CL]。
^Long, Jieyi (15 May 2023). "Large Language Model Guided Tree-of-Thought". arXiv:2305.08291 [cs.AI]。
^Yao, Shunyu; Yu, Dian; Zhao, Jeffrey; Shafran, Izhak; Griffiths, Thomas L.; Cao, Yuan; Narasimhan, Karthik (17 May 2023). "Tree of Thoughts: Deliberate Problem Solving with Large Language Models". arXiv:2305.10601 [cs.CL]。
^Li, Zekun; Peng, Baolin; He, Pengcheng; Galley, Michel; Gao, Jianfeng; Yan, Xifeng (2023). "Guiding Large Language Models via Directional Stimulus Prompting". arXiv:2302.11520 [cs.CL]. The directional stimulus serves as hints or cues for each input query to guide LLMs toward the desired output, such as keywords that the desired summary should include for summarization.
^Zhou, Yongchao; Ioan Muresanu, Andrei; Han, Ziwen; Paster, Keiran; Pitis, Silviu; Chan, Harris; Ba, Jimmy (1 November 2022). "Large Language Models Are Human-Level Prompt Engineers". arXiv:2211.01910 [cs.LG]。
^Zhang, Zhuosheng; Zhang, Aston; Li, Mu; Smola, Alex (1 October 2022). "Automatic Chain of Thought Prompting in Large Language Models". arXiv:2210.03493 [cs.CL]。
^Gal, Rinon; Alaluf, Yuval; Atzmon, Yuval; Patashnik, Or; Bermano, Amit H.; Chechik, Gal; Cohen-Or, Daniel (2022). "An Image is Worth One Word: Personalizing Text-to-Image Generation using Textual Inversion". arXiv:2208.01618. Using only 3-5 images of a user-provided concept, like an object or a style, we learn to represent it through new "words" in the embedding space of a frozen text-to-image model.
^Li, Xiang Lisa; Liang, Percy (2021). “Prefix-Tuning: Optimizing Continuous Prompts for Generation”. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). pp. 4582-4597. doi:10.18653/V1/2021.ACL-LONG.353. ""In this paper, we propose prefix-tuning, a lightweight alternative to fine-tuning... Prefix-tuning draws inspiration from prompting""
^Lester, Brian; Al-Rfou, Rami; Constant, Noah (2021). “The Power of Scale for Parameter-Efficient Prompt Tuning”. Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing. pp. 3045-3059. arXiv:2104.08691. doi:10.18653/V1/2021.EMNLP-MAIN.243. ""In this work, we explore "prompt tuning," a simple yet effective mechanism for learning "soft prompts"...Unlike the discrete text prompts used by GPT-3, soft prompts are learned through back-propagation""
^Sun, Simeng; Liu, Yang; Iter, Dan; Zhu, Chenguang; Iyyer, Mohit (2023). "How Does In-Context Learning Help Prompt Tuning?". arXiv:2302.11521 [cs.CL]。
^Selvi, Jose (2022年12月5日). “Exploring Prompt Injection Attacks”. research.nccgroup.com. 2023年8月20日閲覧。 “Prompt Injection is a new vulnerability that is affecting some AI/ML models and, in particular, certain types of language models using prompt-based learning”