Work on scaling laws has found that LLMs (LMs) show predictable improvements to overall loss with increased scale (model size, training data, and compute). Here, we present evidence for the claim that LMs may show inverse scaling, or worse task performance with increased scale, e.g., due to flaws in the training objective and data. We present empirical evidence of inverse scaling on 11 datasets collected by running a public contest, the Inverse Scaling Prize, with a substantial prize pool. Through analysis of the datasets, along with other examples found in the literature, we identify four potential causes of inverse scaling: (i) preference to repeat memorized sequences over following in-context instructions, (ii) imitation of undesirable patterns in the training data, (iii) tasks containing an easy distractor task which LMs could focus on, rather than the harder real task, and (iv) correct but misleading few-shot demonstrations of the task. We release the winning datasets at https://inversescaling.com/data to allow for further investigation of inverse scaling. Our tasks have helped drive the discovery of U-shaped and inverted-U scaling trends, where an initial trend reverses, suggesting that scaling trends are less reliable at predicting the behavior of larger-scale models than previously understood. Overall, our results suggest that there are tasks for which increased model scale alone may not lead to progress, and that more careful thought needs to go into the data and objectives for training language models.
We're not able to analyze this paper right now due to high demand.
Please check back later (sorry!).
Generate a detailed summary of this paper with a premium account.
We ran into a problem analyzing this paper.
Anthropic. Introducing Claude. https://www.anthropic.com/index/introducing-claude, 2023. Accessed: 2023-03-15.
Language Models are Few-Shot Learners. In H. Larochelle, M. Ranzato, R. Hadsell, M.F. Balcan, and H. Lin (eds.), Advances in Neural Information Processing Systems, volume 33, pp. 1877–1901. Curran Associates, Inc., 2020. https://proceedings.neurips.cc/paper_files/paper/2020/file/1457c0d6bfcb4967418bfb8ac142f64a-Paper.pdf.
Deep Reinforcement Learning from Human Preferences. In Advances in Neural Information Processing Systems, volume 30. Curran Associates, Inc., 2017. https://proceedings.neurips.cc/paper/2017/file/d5e2c0adad503c91f91df240d0cd4e49-Paper.pdf.
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), pp. 4171–4186, Minneapolis, Minnesota, June 2019. Association for Computational Linguistics. doi: 10.18653/v1/N19-1423. https://aclanthology.org/N19-1423.
Behavioral Biases in the NFL Gambling Market: Overreaction to News and the Recency Bias. SSRN Electronic Journal, 2021. doi: 10.2139/ssrn.3861231. https://doi.org/10.2139/ssrn.3861231.
On the Psychology of Prediction. Psychological Review, 80(4):237–251, July 1973. doi: 10.1037/h0034747. https://doi.org/10.1037/h0034747.
Deduplicating Training Data Mitigates Privacy Risks in Language Models. In ICML, 2022. https://proceedings.mlr.press/v162/kandpal22a/kandpal22a.pdf.
TruthfulQA: Measuring How Models Mimic Human Falsehoods. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pp. 3214–3252, Dublin, Ireland, May 2022. Association for Computational Linguistics. doi: 10.18653/v1/2022.acl-long.229. https://aclanthology.org/2022.acl-long.229.
Inverse Scaling Prize Ideas, Oct 2022. https://ethanperez.net/inverse-scaling-prize-ideas/.
Can a Suit of Armor Conduct Electricity? A New Dataset for Open Book Question Answering. In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pp. 2381–2391, Brussels, Belgium, October-November 2018. Association for Computational Linguistics. doi: 10.18653/v1/D18-1260. https://aclanthology.org/D18-1260.
OpenAI. Model index for researchers. https://platform.openai.com/docs/model-index-for-researchers, 2022. Accessed: 2023-04-06.
The LAMBADA dataset: Word prediction requiring a broad discourse context. In Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pp. 1525–1534, Berlin, Germany, August 2016. Association for Computational Linguistics. doi: 10.18653/v1/P16-1144. https://aclanthology.org/P16-1144.
BBQ: A Hand-Built Bias Benchmark for Question Answering. In Findings of the Association for Computational Linguistics: ACL 2022, pp. 2086–2105, Dublin, Ireland, May 2022. Association for Computational Linguistics. doi: 10.18653/v1/2022.findings-acl.165. https://aclanthology.org/2022.findings-acl.165.
Sundar Pichai. An important next step on our AI journey, 2023. https://blog.google/technology/ai/bard-google-ai-search-updates/.
Counterfactual Story Reasoning and Generation. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pp. 5043–5053, Hong Kong, China, November 2019. Association for Computational Linguistics. doi: 10.18653/v1/D19-1509. https://aclanthology.org/D19-1509.
Introducing ChatGPT. https://openai.com/blog/chatgpt, 2022. Accessed: 2023-03-15.
Process for Adapting Language Models to Society (PALMS) with Values-Targeted Datasets. In M. Ranzato, A. Beygelzimer, Y. Dauphin, P. S. Liang, and J. Wortman Vaughan (eds.), Advances in Neural Information Processing Systems, volume 34, pp. 5861–5873. Curran Associates, Inc., 2021. https://proceedings.neurips.cc/paper/2021/file/2e855f9489df0712b4bd8ea9e2848c5a-Paper.pdf.
Learning to Summarize with Human Feedback. In H. Larochelle, M. Ranzato, R. Hadsell, M.F. Balcan, and H. Lin (eds.), Advances in Neural Information Processing Systems, volume 33, pp. 3008–3021. Curran Associates, Inc., 2020. https://proceedings.neurips.cc/paper_files/paper/2020/file/1f89885d556929e98d3ef9b86448f951-Paper.pdf.
GLUE: A Multi-Task Benchmark and Analysis Platform for Natural Language Understanding. In Proceedings of the 2018 EMNLP Workshop BlackboxNLP: Analyzing and Interpreting Neural Networks for NLP, pp. 353–355, Brussels, Belgium, 2018. Association for Computational Linguistics. doi: 10.18653/v1/W18-5446. http://aclweb.org/anthology/W18-5446.
SuperGLUE: A Stickier Benchmark for General-Purpose Language Understanding Systems. In Advances in Neural Information Processing Systems, volume 32. Curran Associates, Inc., 2019. https://papers.nips.cc/paper_files/paper/2019/hash/4496bf24afe7fab6f046bf4923da8de6-Abstract.html.
Identifying and Mitigating Spurious Correlations for Improving Robustness in NLP Models. In Findings of the Association for Computational Linguistics: NAACL 2022, pp. 1719–1729, Seattle, United States, July 2022. Association for Computational Linguistics. doi: 10.18653/v1/2022.findings-naacl.130. https://aclanthology.org/2022.findings-naacl.130.
Peter Cathcart Wason. Reasoning about a Rule. Quarterly Journal of Experimental Psychology, 20(3):273–281, 1968. doi: 10.1080/14640746808400161. https://journals.sagepub.com/doi/10.1080/14640746808400161.
Chain of Thought Prompting Elicits Reasoning in LLMs. In Alice H. Oh, Alekh Agarwal, Danielle Belgrave, and Kyunghyun Cho (eds.), Advances in Neural Information Processing Systems, 2022c. https://openreview.net/forum?id=_VjQlMeSB_J.