Elamur Research⌕
☰
home / papers / 2022 / Predictability and Surprise in Large Generative Models

Predictability and Surprise in Large Generative Models

Paper · 2022 · Deep Ganguli, Danny Hernandez, Liane Lovitt et al.

Summary

This Anthropic position paper (FAccT '22) examines a tension in large generative models such as GPT-3, Gopher, LaMDA and Megatron-Turing NLG. Scaling laws let developers forecast how loss on a broad training distribution drops as compute, data and parameters increase. That forecastability makes investment in bigger systems look low-risk. Yet nobody can know in advance which specific skills a model will have, what inputs it will meet, or what it will produce. The authors split this thesis into four properties: smooth general scaling, abrupt task-specific scaling, open-ended inputs and open-ended outputs. They support each property with published results, real-world observations and two new experiments on language models. These experiments illustrate harms that arise from unpredictability, including harmful outputs that emerge as models grow. The paper then analyzes the economic, scientific and prestige incentives that push deployment, along with barriers such as cost, talent and missing standards. It also measures a growing gap between academia and industry. The work is significant because it forecasts a spread of actors releasing powerful but poorly understood models, and it proposes concrete policy responses.

Key points

  • The paper identifies four distinguishing features of large generative models: smooth general capability scaling, abrupt specific capability scaling, open-ended inputs/domains, and open-ended outputs.
  • Figure 1 is adapted from earlier scaling-law work (ref. [43]). It shows test loss following power laws in compute (petaflop-days), dataset size (tokens) and number of parameters.
  • Sudden capability jumps are illustrated with three examples from prior literature. In the authors' experience, such jumps affect a minority of tasks but are not rare. They call for systematic study of these jumps, including through mechanistic interpretability.
  • The empirical contributions are two novel experiments on harms from unpredictability, examples of toxic outputs that increase with scale, and a quantitative analysis of the academia–industry divide in large-model development.
  • Recommended interventions include:
    • governance structures that give broader stakeholders a say in deployment decisions;
    • government support for measuring, monitoring and auditing AI systems;
    • funding for evaluation tooling, such as Eleuther's Language Model Evaluation Harness, BIG-bench and HuggingFace's BERTology tools.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🏢AnthropicSan Francisco · United States8 works · 1,908 citations

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksArtificial Intelligence · 17,965 worksSocial Sciences · 1,954 worksSafety Research · 626 works

Cite

Cite as Deep Ganguli, Danny Hernandez, Liane Lovitt, Nova DasSarma, Tom Henighan, Andy Jones, Nicholas Joseph, Jackson Kernion, Ben Mann, Amanda Askell, Yuntao Bai, Anna Chen, Tom Conerly, Dawn Drain, Nelson Elhage, Sheer El Showk, Stanislav Fort, Zac Hatfield-Dodds, Scott Johnston, Shauna Kravec, Neel Nanda, Kamal Ndousse, Catherine Olsson, Daniela Amodei, Dario Amodei, Tom Brown, Jared Kaplan, Sam McCandlish, Chris Olah and Jack Clark. 2022. Predictability and Surprise in Large Generative Models. arXiv:2202.07785.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.1145/3531146.3533229 · arXiv 2202.07785

@misc{ganguli2022predictability,
  title = {{Predictability and Surprise in Large Generative Models}},
  author = {Deep Ganguli, Danny Hernandez, Liane Lovitt, Nova DasSarma, Tom Henighan, Andy Jones, Nicholas Joseph, Jackson Kernion, Ben Mann, Amanda Askell, Yuntao Bai, Anna Chen, Tom Conerly, Dawn Drain, Nelson Elhage, Sheer {El Showk}, Stanislav Fort, Zac Hatfield-Dodds, Scott Johnston, Shauna Kravec, Neel Nanda, Kamal Ndousse, Catherine Olsson, Daniela Amodei, Dario Amodei, Tom Brown, Jared Kaplan, Sam McCandlish, Chris Olah and Jack Clark},
  year = {2022},
  doi = {10.1145/3531146.3533229},
  eprint = {2202.07785},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.1145/3531146.3533229}
}

References

90 references, 34 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Neel Alex, Eli Lifland, Lewis Tunstall, Abhishek Thakur, Pegah Maham, C. Jess Riedel, Emmie Hine, Carolyn Ashurst, Paul Sedille, Alexis Carlier, Michael Noetel, and Andreas Stuhlmüller. 2021. RAFT: A Real-World Few-Shot Text Classification Benchmark. arXiv:2109.14076 [cs] (Nov. 2021). http://arxiv.o… — [arXiv:2109.14076](https://arxiv.org/abs/2109.14076) ↗
  2. Julia Angwin, Jeff Larson, Surya Mattu, and Lauren Kirchner. 2016. Machine bias: There’s software used across the country to predict future criminals. and it’s biased against blacks. https://www.propublica.org/article/machine-bias-risk-assessments-in-criminal-sentencing ↗
  3. Amanda Askell, Yuntao Bai, Anna Chen, Dawn Drain, Deep Ganguli, Tom Henighan, Andy Jones, Nicholas Joseph, Ben Mann, Nova DasSarma, Nelson Elhage, Zac Hatfield-Dodds, Danny Hernandez, Jackson Kernion, Kamal Ndousse, Catherine Olsson, Dario Amodei, Tom Brown, Jack Clark, Sam McCandlish, Chris Olah, a… find
  4. Jacob Austin, Augustus Odena, Maxwell Nye, Maarten Bosma, Henryk Michalewski, David Dohan, Ellen Jiang, Carrie Cai, Michael Terry, Quoc Le, and Charles Sutton. 2021. Program Synthesis with Large Language Models. arXiv:2108.07732 [cs] (Aug. 2021). http://arxiv.org/abs/2108.07732 arXiv: 2108.07732. ↗
  5. Shahar Avin, Haydn Belfield, Miles Brundage, Gretchen Krueger, Jasmine Wang, Adrian Weller, Markus Anderljung, Igor Krawczuk, David Krueger, Jonathan Lebensold, Tegan Maharaj, and Noa Zilberman. 2021. Filling gaps in trustworthy development of AI. Science (Dec. 2021). https://doi.org/10.1126/science… — [doi:10.1126/science.abi7176](https://doi.org/10.1126/science.abi7176) ↗
  6. Michelle Bao, Angela Zhou, Samantha Zottola, Brian Brubach, Sarah Desmarais, Aaron Horowitz, Kristian Lum, and Suresh Venkatasubramanian. 2021. It’s COMPASlicated: The Messy Relationship between RAI Datasets and Algorithmic Fairness Benchmarks. arXiv:2106.05498 [cs] (June 2021). http://arxiv.org/abs… — [arXiv:2106.05498](https://arxiv.org/abs/2106.05498) ↗
  7. Emily M. Bender, Timnit Gebru, Angelina McMillan-Major, and Shmargaret Shmitchell. 2021. On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?. In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency (FAccT ’21). Association for Computing Machinery, Ne… — [doi:10.1145/3442188.3445922](https://doi.org/10.1145/3442188.3445922) find
  8. Su Lin Blodgett, Solon Barocas, Hal Daumé III, and Hanna Wallach. 2020. Language (Technology) is Power: A Critical Survey of "Bias" in NLP. arXiv:2005.14050 [cs] (May 2020). http://arxiv.org/abs/2005.14050 arXiv: 2005.14050. ↗
  9. Rishi Bommasani, Drew A. Hudson, Ehsan Adeli, Russ Altman, Simran Arora, Sydney von Arx, Michael S. Bernstein, Jeannette Bohg, Antoine Bosselut, Emma Brunskill, Erik Brynjolfsson, Shyamal Buch, Dallas Card, Rodrigo Castellon, Niladri Chatterji, Annie Chen, Kathleen Creel, Jared Quincy Davis, Dora De… — On the Opportunities and Risks of Foundation Models
  10. Sebastian Borgeaud, Arthur Mensch, Jordan Hoffmann, Trevor Cai, Eliza Rutherford, Katie Millican, George van den Driessche, Jean-Baptiste Lespiau, Bogdan Damoc, Aidan Clark, Diego de Las Casas, Aurelia Guy, Jacob Menick, Roman Ring, Tom Hennigan, Saffron Huang, Loren Maggiore, Chris Jones, Albin Cas… find
  11. Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert-Voss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel Ziegler, Jeffrey Wu, Clemens Winter, Chri… — Language Models are Few-Shot Learners
  12. Miles Brundage, Shahar Avin, Jasmine Wang, Haydn Belfield, Gretchen Krueger, Gillian Hadfield, Heidy Khlaaf, Jingying Yang, Helen Toner, Ruth Fong, Tegan Maharaj, Pang Wei Koh, Sara Hooker, Jade Leung, Andrew Trask, Emma Bluemke, Jonathan Lebensold, Cullen O’Keefe, Mark Koren, Théo Ryffel, J. B. Rub… find
  13. Ben Buchanan, Andrew Lohn, Micha Musser, and Katerina Sedova. 2021. Truth, Lies, and Automation. https://cset.georgetown.edu/publication/truth-lies-and-automation/ ↗
  14. Nicholas Carlini, Florian Tramer, Eric Wallace, Matthew Jagielski, Ariel Herbert-Voss, Katherine Lee, Adam Roberts, Tom Brown, Dawn Song, Ulfar Erlingsson, Alina Oprea, and Colin Raffel. 2021. Extracting Training Data from Large Language Models. arXiv:2012.07805 [cs] (June 2021). http://arxiv.org/ab… ↗
  15. Edwin Cartlidge. 2019. Square Kilometre Array hit with further cost hike and delay. Physics World (Aug. 2019). https://physicsworld.com/a/square-kilometre-array-hit-with-further-cost-hike-and-delay/ ↗
  16. Mark Chen, Jerry Tworek, Heewoo Jun, Qiming Yuan, Henrique Ponde de Oliveira Pinto, Jared Kaplan, Harri Edwards, Yuri Burda, Nicholas Joseph, Greg Brockman, Alex Ray, Raul Puri, Gretchen Krueger, Michael Petrov, Heidy Khlaaf, Girish Sastry, Pamela Mishkin, Brooke Chan, Scott Gray, Nick Ryder, Mikhai… — Evaluating Large Language Models Trained on Code
  17. Kevin Clark, Urvashi Khandelwal, Omer Levy, and Christopher D. Manning. 2019. What Does BERT Look At? An Analysis of BERT’s Attention. arXiv:1906.04341 [cs] (June 2019). http://arxiv.org/abs/1906.04341 arXiv: 1906.04341. ↗
  18. Kaleigh Clary, Emma Tosch, John Foley, and David Jensen. 2019. Let’s Play Again: Variability of Deep Reinforcement Learning Agents in Atari Environments. arXiv:1904.06312 [cs, stat] (April 2019). http://arxiv.org/abs/1904.06312 arXiv: 1904.06312. — [arXiv:1904.06312](https://arxiv.org/abs/1904.06312) ↗
  19. Kate Crawford. 2021. Atlas of AI. Yale University Press. https://yalebooks.yale.edu/book/9780300209570/atlas-ai ↗
  20. Emily Denton, Alex Hanna, Razvan Amironesei, Andrew Smart, Hilary Nicole, and Morgan Klaus Scheuerman. 2020. Bringing the People Back In: Contesting Benchmark Machine Learning Datasets. arXiv:2007.07399 [cs] (July 2020). http://arxiv.org/abs/2007.07399 arXiv: 2007.07399. — [arXiv:2007.07399](https://arxiv.org/abs/2007.07399) ↗
  21. Emily Dinan, Gavin Abercrombie, A. Stevie Bergman, Shannon Spruit, Dirk Hovy, Y.-Lan Boureau, and Verena Rieser. 2021. Anticipating Safety Issues in E2E Conversational AI: Framework and Tooling. arXiv:2107.03451 [cs] (July 2021). http://arxiv.org/abs/2107.03451 arXiv: 2107.03451. — [arXiv:2107.03451](https://arxiv.org/abs/2107.03451) ↗
  22. Julia Dressel and Hany Farid. 2018. The accuracy, fairness, and limits of predicting recidivism. Science Advances (Jan. 2018). https://doi.org/10.1126/ sciadv.aao5580 Publisher: American Association for the Advancement of Science. find
  23. Jasha Droppo and Oguz Elibol. 2021. Scaling Laws for Acoustic Models. arXiv:2106.09488 [cs, eess] (June 2021). http://arxiv.org/abs/2106.09488 arXiv: 2106.09488. — [arXiv:2106.09488](https://arxiv.org/abs/2106.09488) ↗
  24. Nan Du, Yanping Huang, Andrew M. Dai, Simon Tong, Dmitry Lepikhin, Yuanzhong Xu, Maxim Krikun, Yanqi Zhou, Adams Wei Yu, Orhan Firat, Barret Zoph, Liam Fedus, Maarten Bosma, Zongwei Zhou, Tao Wang, Yu Emma Wang, Kellie Webster, Marie Pellat, Kevin Robinson, Kathy Meier-Hellstern, Toju Duke, Lucas Di… — GLaM: Efficient Scaling of Language Models with Mixture-of-Experts
  25. Nelson Elhage, Neel Nanda, Catherine Olsson, Tom Henighan, Nicholas Joseph, Ben Mann, Amanda Askell, Yuntao Bai, Anna Chen, Tom Conerly, Nova DasSarma, Dawn Drain, Deep Ganguli, Zac Hatfield-Dodds, Danny Hernandez, Andy Jones, Jackson Kernion, Liane Lovitt, Kamal Ndousse, Dario Amodei, Tom Brown, Ja… find
  26. Sorelle A. Friedler, Carlos Scheidegger, and Suresh Venkatasubramanian. 2016. On the (im)possibility of fairness. arXiv:1609.07236 [cs, stat] (Sept. 2016). http://arxiv.org/abs/1609.07236 arXiv: 1609.07236. — [arXiv:1609.07236](https://arxiv.org/abs/1609.07236) ↗
  27. Leo Gao, Jonathan Tow, Stella Biderman, Sid Black, Anthony DiPofi, Charles Foster, Laurence Golding, Jeffrey Hsu, Kyle McDonell, Niklas Muennighoff, Jason Phang, Laria Reynolds, Eric Tang, Anish Thite, Ben Wang, Kevin Wang, and Andy Zou. 2021. A framework for few-shot language model evaluation. http… — [doi:10.5281/zenodo.5371628](https://doi.org/10.5281/zenodo.5371628) find
  28. Timnit Gebru, Jamie Morgenstern, Briana Vecchione, Jennifer Wortman Vaughan, Hanna Wallach, Hal Daumé III, and Kate Crawford. 2021. Datasheets for Datasets. arXiv:1803.09010 [cs] (Dec. 2021). http://arxiv.org/abs/1803.09010 arXiv: 1803.09010. ↗
  29. Samuel Gehman, Suchin Gururangan, Maarten Sap, Yejin Choi, and Noah A. Smith. 2020. RealToxicityPrompts: Evaluating Neural Toxic Degeneration in Language Models. ArXiv abs/2009.11462 (2020).
  30. Mary Gray and Siddharth Suri. 2019. Ghost Work. Mariner Books. https://ghostwork.info/ ↗
  31. Soyeon Caren Han, Taejun Lim, Siqu Long, Bernd Burgstaller, and Josiah Poon. 2021. GLocal-K: Global and Local Kernels for Recommender Systems. (Aug. 2021). https://doi.org/10.1145/3459637.3482112 — [doi:10.1145/3459637.3482112](https://doi.org/10.1145/3459637.3482112) ↗
  32. Laura Hanu and Unitary team. 2020. Detoxify. Published: Github. https://github.com/unitaryai/detoxify. ↗
  33. F. Maxwell Harper and Joseph A. Konstan. 2015. The MovieLens Datasets: History and Context. ACM Transactions on Interactive Intelligent Systems 5, 4 (Dec. 2015), 19:1–19:19. https://doi.org/10.1145/2827872 — [doi:10.1145/2827872](https://doi.org/10.1145/2827872) ↗
  34. Dan Hendrycks, Collin Burns, Steven Basart, Andy Zou, Mantas Mazeika, Dawn Song, and Jacob Steinhardt. 2021. Measuring Massive Multitask Language Understanding. arXiv:2009.03300 [cs] (Jan. 2021). http://arxiv.org/abs/2009.03300 arXiv: 2009.03300. ↗
  35. Dan Hendrycks, Nicholas Carlini, John Schulman, and Jacob Steinhardt. 2021. Unsolved Problems in ML Safety. arXiv:2109.13916 [cs] (Dec. 2021). http://arxiv.org/abs/2109.13916 arXiv: 2109.13916. ↗
  36. Dan Hendrycks, Kevin Zhao, Steven Basart, Jacob Steinhardt, and Dawn Song. 2021. Natural Adversarial Examples. arXiv:1907.07174 [cs, stat] (March 2021). http://arxiv.org/abs/1907.07174 arXiv: 1907.07174. ↗
  37. Tom Henighan, Jared Kaplan, Mor Katz, Mark Chen, Christopher Hesse, Jacob Jackson, Heewoo Jun, Tom B. Brown, Prafulla Dhariwal, Scott Gray, Chris Hallacy, Benjamin Mann, Alec Radford, Aditya Ramesh, Nick Ryder, Daniel M. Ziegler, John Schulman, Dario Amodei, and Sam McCandlish. 2020. Scaling Laws fo…
  38. Danny Hernandez, Jared Kaplan, Tom Henighan, and Sam McCandlish. 2021. Scaling Laws for Transfer. arXiv:2102.01293 [cs] (Feb. 2021). http://arxiv.org/abs/2102.01293 arXiv: 2102.01293. ↗
  39. Joel Hestness, Sharan Narang, Newsha Ardalani, Gregory Diamos, Heewoo Jun, Hassan Kianinejad, Md Mostofa Ali Patwary, Yang Yang, and Yanqi Zhou. 2017. Deep Learning Scaling is Predictable, Empirically. arXiv:1712.00409 [cs, stat] (Dec. 2017). http://arxiv.org/abs/1712.00409 arXiv: 1712.00409. ↗
  40. Daniel Ho, Jennifer King, Russell Wald, and Christopher Wan. 2021. Building a National AI Research Resource. White Paper. Stanford University. https://hai.stanford.edu/white-paper-building-national-ai-research-resource ↗
  41. Erik Hoel. 2021. Big Tech is replacing human artists with AI. https://erikhoel.substack.com/p/big-tech-is-replacing-human-artists ↗
  42. AI Now Institute. 2021. Democratize AI? How the proposed National AI Research Resource falls short. https://medium.com/@AINowInstitute/ democratize-ai-how-the-proposed-national-ai-research-resource-falls-short-96ae5f67ccfa ↗
  43. Jared Kaplan, Sam McCandlish, Tom Henighan, Tom B. Brown, Benjamin Chess, Rewon Child, Scott Gray, Alec Radford, Jeffrey Wu, and Dario Amodei. 2020. Scaling Laws for Neural Language Models. arXiv:2001.08361 [cs, stat] (Jan. 2020). http://arxiv.org/abs/2001.08361 arXiv: 2001.08361. ↗
  44. Zachary Kenton, Tom Everitt, Laura Weidinger, Iason Gabriel, Vladimir Mikulik, and Geoffrey Irving. 2021. Alignment of Language Agents. arXiv:2103.14659 [cs] (March 2021). http://arxiv.org/abs/2103.14659 arXiv: 2103.14659. ↗
  45. Josh Kenway, Francois Camille, Sasha Costanza-Chock, Deborah Raji, Inioluwa, and Joy Buolamwini. 2022. Bug Bounties For Algorithmic Harms? Technical Report. Algorithmic Justice League. https://www.ajl.org/bugs ↗
  46. Boseop Kim, HyoungSeok Kim, Sang-Woo Lee, Gichang Lee, Donghyun Kwak, Dong Hyeon Jeon, Sunghyun Park, Sungju Kim, Seonhoon Kim, Dongpil Seo, Heungsub Lee, Minyoung Jeong, Sungjae Lee, Minsub Kim, Suk Hyun Ko, Seokhun Kim, Taeyong Park, Jinuk Kim, Soyoung Kang, Na-Hyeon Ryu, Kang Min Yoo, Minsuk Chan… find
  47. Alex Knapp. 2012. How Much Does It Cost To Find A Higgs Boson? Forbes (June 2012). https://www.forbes.com/sites/alexknapp/2012/07/05/how-much-does-it-cost-to-find-a-higgs-boson/ Section: Tech. ↗
  48. Alex Krizhevsky, Ilya Sutskever, and Geoffrey E Hinton. 2012. ImageNet Classification with Deep Convolutional Neural Networks. In Advances in Neural Information Processing Systems, F. Pereira, C. J. C. Burges, L. Bottou, and K. Q. Weinberger (Eds.), Vol. 25. Curran Associates, Inc. https://proceedin…
  49. Connor Leahy. 2022. Announcing GPT-NeoX-20B. https://blog.eleuther.ai/announcing-20b/ ↗
  50. Chuan Li. 2020. OpenAI’s GPT-3 Language Model: A Technical Overview. https://lambdalabs.com/blog/demystifying-gpt-3/ ↗
  51. Opher Lieber, Or Sharir, Barak Lenz, and Yoav Shoham. 2021. Jurassic-1: Technical Details And Evaluation. Technical Report. AI21 Labs. find
  52. Andrew Lohn and Micha Musser. 2022. AI and Compute: How Much Longer Can Computing Power Drive Artificial Intelligence Progress? Technical Report. Center for Security and Emerging Technology. https://doi.org/10.51593/2021CA009 — [doi:10.51593/2021ca009](https://doi.org/10.51593/2021ca009) ↗
  53. Margaret Mitchell, Simone Wu, Andrew Zaldivar, Parker Barnes, Lucy Vasserman, Ben Hutchinson, Elena Spitzer, Inioluwa Deborah Raji, and Timnit Gebru. 2019. Model Cards for Model Reporting. Proceedings of the Conference on Fairness, Accountability, and Transparency (Jan. 2019), 220–229. https://doi.o… ↗
  54. Shakir Mohamed, Marie-Therese Png, and William Isaac. 2020. Decolonial AI: Decolonial Theory as Sociotechnical Foresight in Artificial Intelligence. Philosophy & Technology 33 (Dec. 2020). https://doi.org/10.1007/s13347-020-00405-8 — [doi:10.1007/s13347-020-00405-8](https://doi.org/10.1007/s13347-020-00405-8) ↗
  55. Nick Walton [@nickwalton00]. 2020. I’ve noticed a number of people using AI Dungeon to test GPT-3’s abilities. While it’s a great way to see how GPT-3 can power an interesting application. It’s a poor test of GPT-3’s abilities in general. The first generation of any custom prompt is actually GPT-2.… find
  56. David Patterson, Joseph Gonzalez, Quoc Le, Chen Liang, Lluis-Miquel Munguia, Daniel Rothchild, David So, Maud Texier, and Jeff Dean. 2021. Carbon Emissions and Large Neural Network Training. arXiv:2104.10350 [cs] (April 2021). http://arxiv.org/abs/2104.10350 arXiv: 2104.10350. ↗
  57. Ethan Perez, Saffron Huang, Francis Song, Trevor Cai, Roman Ring, John Aslanides, Amelia Glaese, Nat McAleese, and Geoffrey Irving. 2022. Red Teaming Language Models with Language Models. arXiv:2202.03286 [cs] (Feb. 2022). http://arxiv.org/abs/2202.03286 arXiv: 2202.03286. ↗
  58. Alethea Power, Yuri Burda, Harri Edwards, Igor Babuschkin, and Vedant Misra. 2022. Grokking: Generalization Beyond Overfitting on Small Algorithmic Datasets. arXiv:2201.02177 [cs] (Jan. 2022). http://arxiv.org/abs/2201.02177 arXiv: 2201.02177. ↗
  59. Shrimai Prabhumoye, Rafal Kocielnik, Mohammad Shoeybi, Anima Anandkumar, and Bryan Catanzaro. 2021. Few-shot Instruction Prompts for Pretrained Language Models to Detect Social Biases. arXiv:2112.07868 [cs] (Dec. 2021). http://arxiv.org/abs/2112.07868 arXiv: 2112.07868. — [arXiv:2112.07868](https://arxiv.org/abs/2112.07868) ↗
  60. Gabriele Prato, Simon Guiroy, Ethan Caballero, Irina Rish, and Sarath Chandar. 2021. Scaling Laws for the Few-Shot Adaptation of Pre-trained Image Classifiers. arXiv:2110.06990 [cs] (Oct. 2021). http://arxiv.org/abs/2110.06990 arXiv: 2110.06990. — [arXiv:2110.06990](https://arxiv.org/abs/2110.06990) ↗
  61. Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, and Ilya Sutskever. 2021. Learning Transferable Visual Models From Natural Language Supervision. arXiv:2103.00020 [cs] (Feb. 2021). ht…
  62. Jack W. Rae, Sebastian Borgeaud, Trevor Cai, Katie Millican, Jordan Hoffmann, Francis Song, John Aslanides, Sarah Henderson, Roman Ring, Susannah Young, Eliza Rutherford, Tom Hennigan, Jacob Menick, Albin Cassirer, Richard Powell, George van den Driessche, Lisa Anne Hendricks, Maribeth Rauh, Po-Sen… find
  63. Inioluwa Deborah Raji, Emily M. Bender, Amandalynne Paullada, Emily Denton, and Alex Hanna. 2021. AI and the Everything in the Whole Wide World Benchmark. arXiv:2111.15366 [cs] (Nov. 2021). http://arxiv.org/abs/2111.15366 arXiv: 2111.15366. ↗
  64. Inioluwa Deborah Raji and Joy Buolamwini. 2019. Actionable Auditing: Investigating the Impact of Publicly Naming Biased Performance Results of Commercial AI Products. In Proceedings of the 2019 AAAI/ACM Conference on AI, Ethics, and Society (AIES ’19). Association for Computing Machinery, New York,… — [doi:10.1145/3306618.3314244](https://doi.org/10.1145/3306618.3314244) find
  65. Inioluwa Deborah Raji, Andrew Smart, Rebecca N. White, Margaret Mitchell, Timnit Gebru, Ben Hutchinson, Jamila Smith-Loud, Daniel Theron, and Parker Barnes. 2020. Closing the AI accountability gap: defining an end-to-end framework for internal algorithmic auditing. In Proceedings of the 2020 Confere… — [doi:10.1145/3351095.3372873](https://doi.org/10.1145/3351095.3372873) find
  66. Jonathan S. Rosenfeld, Amir Rosenfeld, Yonatan Belinkov, and Nir Shavit. 2019. A Constructive Prediction of the Generalization Error Across Scales. arXiv:1909.12673 [cs, stat] (Dec. 2019). http://arxiv.org/abs/1909.12673 arXiv: 1909.12673. ↗
  67. Cynthia Rudin, Caroline Wang, and Beau Coker. 2020. The Age of Secrecy and Unfairness in Recidivism Prediction. Harvard Data Science Review 2, 1 (March 2020). https://doi.org/10.1162/99608f92.6ed64b30 ↗
  68. Victor Sanh, Albert Webson, Colin Raffel, Stephen H. Bach, Lintang Sutawika, Zaid Alyafeai, Antoine Chaffin, Arnaud Stiegler, Teven Le Scao, Arun Raja, Manan Dey, M. Saiful Bari, Canwen Xu, Urmish Thakker, Shanya Sharma Sharma, Eliza Szczechla, Taewoon Kim, Gunjan Chhablani, Nihal Nayak, Debajyoti D… — Multitask Prompted Training Enables Zero-Shot Task Generalization
  69. Roy Schwartz, Jesse Dodge, Noah A. Smith, and Oren Etzioni. 2020. Green AI. Commun. ACM 63, 12 (Nov. 2020), 54–63. https://doi.org/10.1145/ 3381831
  70. Jaime Sevilla, Pablo Villalobos, Juan Felipe Cerón, Matthew Burtell, Lennart Heim, Amogh B Nanjajjar, Anson Ho, Tamay Besiroglu, Marius Hobbhahn, and Jean-Stanislas Denain. 2021. Parameter, Compute and Data Trends in Machine Learning. https://docs.google.com/spreadsheets/d/ 1AAIebjNsnJj_uKALHbXNfn3_… ↗
  71. Tom Simonite. 2021. What Really Happened When Google Ousted Timnit Gebru. Wired (June 2021). https://www.wired.com/story/google-timnit-gebru-ai-what-really-happened/ Section: tags. ↗
  72. Shaden Smith, Mostofa Patwary, Brandon Norick, Patrick LeGresley, Samyam Rajbhandari, Jared Casper, Zhun Liu, Shrimai Prabhumoye, George Zerveas, Vijay Korthikanti, Elton Zhang, Rewon Child, Reza Yazdani Aminabadi, Julie Bernauer, Xia Song, Mohammad Shoeybi, Yuxiong He, Michael Houston, Saurabh Tiwa… — Using DeepSpeed and Megatron to Train Megatron-Turing NLG 530B, A Large-Scale Generative Language Model
  73. Irene Solaiman and Christy Dennison. 2021. Process for Adapting Language Models to Society (PALMS) with Values-Targeted Datasets. arXiv:2106.10328 [cs] (Nov. 2021). http://arxiv.org/abs/2106.10328 arXiv: 2106.10328. ↗
  74. Emma Strubell, Ananya Ganesh, and Andrew McCallum. 2019. Energy and Policy Considerations for Deep Learning in NLP. arXiv:1906.02243 [cs] (June 2019). http://arxiv.org/abs/1906.02243 arXiv: 1906.02243. ↗
  75. Alex Tamkin, Miles Brundage, Jack Clark, and Deep Ganguli. 2021. Understanding the Capabilities, Limitations, and Societal Impact of Large Language Models. arXiv:2102.02503 [cs] (Feb. 2021). http://arxiv.org/abs/2102.02503 arXiv: 2102.02503. ↗
  76. Latitude Team. 2020. AI Dungeon: Dragon Model Upgrade. https://aidungeon.medium.com/ai-dungeon-dragon-model-upgrade-7e8ea579abfe ↗
  77. Will Thomas. 2020. Flagship Neutrino Project Working to Keep Costs Within Cap. https://www.aip.org/fyi/2020/flagship-neutrino-project-working-keep-costs-within-cap Publisher: American Institute of Physics. ↗
  78. Romal Thoppilan, Daniel De Freitas, Jamie Hall, Noam Shazeer, Apoorv Kulshreshtha, Heng-Tze Cheng, Alicia Jin, Taylor Bos, Leslie Baker, Yu Du, YaGuang Li, Hongrae Lee, Huaixiu Steven Zheng, Amin Ghafouri, Marcelo Menegali, Yanping Huang, Maxim Krikun, Dmitry Lepikhin, James Qin, Dehao Chen, Yuanzho… — LaMDA: Language Models for Dialog Applications
  79. Ted Underwood. 2021. Science fiction hasn’t prepared us to imagine machine learning. https://tedunderwood.com/2021/02/02/why-sf-hasnt-prepared-us-to-imagine-machine-learning/ ↗
  80. Ben Wang and Aran Komatsuzaki. 2021. GPT-J-6B: A 6 Billion Parameter Autoregressive Language Model. https://github.com/kingoflolz/mesh-transformer-jax ↗
  81. Shuohuan Wang, Yu Sun, Yang Xiang, Zhihua Wu, Siyu Ding, Weibao Gong, Shikun Feng, Junyuan Shang, Yanbin Zhao, Chao Pang, Jiaxiang Liu, Xuyi Chen, Yuxiang Lu, Weixin Liu, Xi Wang, Yangfan Bai, Qiuliang Chen, Li Zhao, Shiyong Li, Peng Sun, Dianhai Yu, Yanjun Ma, Hao Tian, Hua Wu, Tian Wu, Wei Zeng, G… — ERNIE 3.0 Titan: Exploring Larger-scale Knowledge Enhanced Pre-training for Language Understanding and Generation
  82. Jason Wei, Maarten Bosma, Vincent Y. Zhao, Kelvin Guu, Adams Wei Yu, Brian Lester, Nan Du, Andrew M. Dai, and Quoc V. Le. 2021. Finetuned Language Models Are Zero-Shot Learners. arXiv:2109.01652 [cs] (Dec. 2021). http://arxiv.org/abs/2109.01652 arXiv: 2109.01652. ↗
  83. Laura Weidinger, John Mellor, Maribeth Rauh, Conor Griffin, Jonathan Uesato, Po-Sen Huang, Myra Cheng, Mia Glaese, Borja Balle, Atoosa Kasirzadeh, Zac Kenton, Sasha Brown, Will Hawkins, Tom Stepleton, Courtney Biles, Abeba Birhane, Julia Haas, Laura Rimell, Lisa Anne Hendricks, William Isaac, Sean L… — Ethical and social risks of harm from Language Models
  84. Johannes Welbl, Amelia Glaese, Jonathan Uesato, Sumanth Dathathri, John Mellor, Lisa Anne Hendricks, Kirsty Anderson, Pushmeet Kohli, Ben Coppin, and Po-Sen Huang. 2021. Challenges in Detoxifying Language Models. In Findings of the Association for Computational Linguistics: EMNLP 2021. Association f…
  85. Jess Whittlestone and Jack Clark. 2021. Why and How Governments Should Monitor AI Development. arXiv:2108.12427 [cs] (Aug. 2021). http://arxiv.org/abs/2108.12427 arXiv: 2108.12427. — [arXiv:2108.12427](https://arxiv.org/abs/2108.12427) ↗
  86. M. J. Wolf, K. W. Miller, and F. S. Grodzinsky. 2017. Why We Should Have Seen That Coming: Comments on Microsoft’s Tay “Experiment,” and Wider Implications. The ORBIT Journal 1, 2 (Jan. 2017), 1–12. https://doi.org/10.29297/orbit.v1i2.49 — [doi:10.29297/orbit.v1i2.49](https://doi.org/10.29297/orbit.v1i2.49) ↗
  87. Shaohua Wu, Xudong Zhao, Tong Yu, Rongguo Zhang, Chong Shen, Hongli Liu, Feng Li, Hong Zhu, Jiangang Luo, Liang Xu, and Xuanwei Zhang. 2021. Yuan 1.0: Large-Scale Pre-trained Language Model in Zero-Shot and Few-Shot Learning. arXiv:2110.04725 [cs] (Oct. 2021). http://arxiv.org/abs/2110.04725 arXiv:… ↗
  88. Jing Xu, Da Ju, Margaret Li, Y.-Lan Boureau, Jason Weston, and Emily Dinan. 2021. Bot-Adversarial Dialogue for Safe Conversational Agents. In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, NAACL-HLT 2021…
  89. Wei Zeng, Xiaozhe Ren, Teng Su, Hui Wang, Yi Liao, Zhiwei Wang, Xin Jiang, ZhenZhang Yang, Kaisheng Wang, Xiaoda Zhang, Chen Li, Ziyan Gong, Yifan Yao, Xinjing Huang, Jun Wang, Jianfeng Yu, Qi Guo, Yue Yu, Yan Zhang, Jin Wang, Hengtao Tao, Dasen Yan, Zexuan Yi, Fang Peng, Fangqing Jiang, Han Zhang,… — PanGu-$\alpha$: Large-scale Autoregressive Pretrained Chinese Language Models with Auto-parallel Computation
  90. Donglin Zhuang, Xingyao Zhang, Shuaiwen Leon Song, and Sara Hooker. 2021. Randomness In Neural Network Training: Characterizing The Impact of Tooling. arXiv:2106.11872 [cs] (June 2021). http://arxiv.org/abs/2106.11872 arXiv: 2106.11872. ↗

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/ganguli-2022-predictability-surprise-large-generative-models/