Elamur Research⌕
☰
home / papers / 2023 / Mixture-of-Experts Meets Instruction Tuning:A Winning Combination for Large Language Models

Mixture-of-Experts Meets Instruction Tuning:A Winning Combination for Large Language Models

Paper · 2023 · Sheng Shen, Le Hou, Yanqi Zhou et al.

Summary

Sparse Mixture-of-Experts (MoE) layers let a language model hold far more parameters while keeping per-token compute roughly fixed. However, MoE models tend to transfer poorly when finetuned directly on single downstream tasks. The authors test whether instruction tuning changes this. They build FLAN-MoE, a family of encoder-decoder models in which every second Transformer feed-forward block is replaced by a routed pool of experts. The variants are Switch, GShard, Expert Choice and ST-MoE, each sending a token to one or two experts. These models are instruction-tuned on the Flan collection of 1,836 tasks. They are compared with compute-matched dense baselines (T5/Flan-T5, PaLM/Flan-PaLM) in three regimes: plain task finetuning, instruction tuning with zero/few-shot evaluation, and instruction tuning followed by task finetuning. Without instruction tuning, MoE trails dense models. With it, MoE gains much more and overtakes them. The largest model, FLAN-ST 32B, beats FLAN-PaLM 62B while using about a third of the FLOPs per token. This positions instruction tuning as a key enabler for efficient sparse LLMs. Multilingual performance stays weak, which matters for non-English work such as Ukrainian.

Key points

  • Training and evaluation data: The instruction mixture combines Muffin, T0-SF, NIV2 (1,554 tasks) and CoT (9 reasoning tasks), for 1,836 tasks in total. Held-out evaluation uses MMLU (57 tasks) and BBH (23 tasks), plus reasoning and QA suites, scored by few-shot exact match.
  • Headline result: FLAN-ST 32B has 259B total parameters and uses 32.1G FLOPs/token. It reaches a normalized average of 63.6, compared with 57.6 for FLAN-PaLM 62B at 91.6G FLOPs/token, and scores 65.4 on MMLU-direct.
  • MoE gains more from instruction tuning: At an equal 13.9G FLOPs/token, Switch XXL rises by 36.4 points (17.8 → 54.2), while T5 XXL rises by 24.0 (27.7 → 51.7). Before instruction tuning, MoE normalized averages (about 17–20) sit below dense T5 (about 26–28).
  • Scaling and recipe: MoE performance scales better with the number of instruction tasks than with the number of experts, and expert usage improves after instruction tuning. The recipe uses a prefix LM objective, 2,048 input and 512 output tokens, dropout 0.05, expert dropout 0.2 and learning rate 1e-4.
  • Limitation: Tuning on English-only instructions hurts multilinguality. FLAN-ST 32B scores only 15.5% on MGSM and 25.1% on TyDiQA, roughly matching vanilla PaLM 62B (18.2% MGSM) and PaLM 8B (25.0% TyDiQA).

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

🎓University of California, BerkeleyBerkeley · United States127 works · 212,523 citations🎓Massachusetts Institute of TechnologyCambridge · United States159 works · 65,157 citations🎓The University of Texas at AustinAustin · United States72 works · 55,722 citations🎓University of Massachusetts AmherstAmherst Center · United States67 works · 64,591 citations

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 9,086 worksArtificial Intelligence · 8,889 worksArts and Humanities · 9,296 worksLanguage and Linguistics · 9,206 works

🧭 Topics 1Research subjects and languages the work is about.Language models & pretraining · 1,343 works

Cite

Cite as Sheng Shen, Le Hou, Yanqi Zhou, Nan Du, Shayne Longpre, Jason Wei, Hyung Won Chung, Barret Zoph, William Fedus, Xinyun Chen, Tu Vu, Yuexin Wu, Wuyang Chen, Albert Webson, Yunxuan Li, Vincent Zhao, Hongkun Yu, Kurt Keutzer, Trevor Darrell and Denny Zhou. 2023. Mixture-of-Experts Meets Instruction Tuning:A Winning Combination for Large Language Models. arXiv:2305.14705.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.48550/arxiv.2305.14705 · arXiv 2305.14705

@misc{shen2023mixture,
  title = {{Mixture-of-Experts Meets Instruction Tuning:A Winning Combination for Large Language Models}},
  author = {Sheng Shen, Le Hou, Yanqi Zhou, Nan Du, Shayne Longpre, Jason Wei, Hyung Won Chung, Barret Zoph, William Fedus, Xinyun Chen, Tu Vu, Yuexin Wu, Wuyang Chen, Albert Webson, Yunxuan Li, Vincent Zhao, Hongkun Yu, Kurt Keutzer, Trevor Darrell and Denny Zhou},
  year = {2023},
  doi = {10.48550/arxiv.2305.14705},
  eprint = {2305.14705},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2305.14705}
}

References

57 references, 28 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Vamsi Aribandi, Yi Tay, Tal Schuster, Jinfeng Rao, Huaixiu Steven Zheng, Sanket Vaibhav Mehta, Honglei Zhuang, Vinh Q Tran, Dara Bahri, Jianmo Ni, et al. Ext5: Towards extreme multi-task scaling for transfer learning. arXiv preprint arXiv:2111.10952, 2021. find
  2. Yuntao Bai, Andy Jones, Kamal Ndousse, Amanda Askell, Anna Chen, Nova DasSarma, Dawn Drain, Stanislav Fort, Deep Ganguli, Tom Henighan, et al. Training a helpful and harmless assistant with reinforcement learning from human feedback. arXiv preprint arXiv:2204.05862, 2022. find
  3. Yuntao Bai, Saurav Kadavath, Sandipan Kundu, Amanda Askell, Jackson Kernion, Andy Jones, Anna Chen, Anna Goldie, Azalia Mirhoseini, Cameron McKinnon, et al. Constitutional ai: Harmlessness from ai feedback. arXiv preprint arXiv:2212.08073, 2022.
  4. Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, et al. Scaling instruction-finetuned language models. arXiv preprint arXiv:2210.11416, 2022.
  5. Aidan Clark, Diego De Las Casas, Aurelia Guy, Arthur Mensch, Michela Paganini, Jordan Hoffmann, Bogdan Damoc, Blake Hechtman, Trevor Cai, Sebastian Borgeaud, et al. Unified scaling laws for routed language models. In ICML, pages 4057–4086. PMLR, 2022. find
  6. Christopher Clark, Kenton Lee, Ming-Wei Chang, Tom Kwiatkowski, Michael Collins, and Kristina Toutanova. Boolq: Exploring the surprising difficulty of natural yes/no questions. arXiv preprint arXiv:1905.10044, 2019.
  7. Peter Clark, Isaac Cowhey, Oren Etzioni, Tushar Khot, Ashish Sabharwal, Carissa Schoenick, and Oyvind Tafjord. Think you have solved question answering? try arc, the ai2 reasoning challenge. arXiv preprint arXiv:1803.05457, 2018. find
  8. Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Mark Chen, Heewoo Jun, Lukasz Kaiser, Matthias Plappert, Jerry Tworek, Jacob Hilton, Reiichiro Nakano, et al. Training verifiers to solve math word problems. arXiv preprint arXiv:2110.14168, 2021.
  9. Wenliang Dai, Junnan Li, Dongxu Li, Anthony Meng Huat Tiong, Junqi Zhao, Weisheng Wang, Boyang Li, Pascale Fung, and Steven Hoi. Instructblip: Towards general-purpose vision-language models with instruction tuning. arXiv preprint arXiv:2305.06500, 2023.
  10. Nan Du, Yanping Huang, Andrew M Dai, Simon Tong, Dmitry Lepikhin, Yuanzhong Xu, Maxim Krikun, Yanqi Zhou, Adams Wei Yu, Orhan Firat, et al. Glam: Efficient scaling of language models with mixture-of-experts. In ICML, pages 5547–5569. PMLR, 2022.
  11. David Eigen, Marc’Aurelio Ranzato, and Ilya Sutskever. Learning factored representations in a deep mixture of experts. arXiv preprint arXiv:1312.4314, 2013. — [arXiv:1312.4314](https://arxiv.org/abs/1312.4314) find
  12. William Fedus, Barret Zoph, and Noam Shazeer. Switch transformers: Scaling to trillion parameter models with simple and efficient sparsity. CoRR, abs/2101.03961, 2021. find
  13. Mor Geva, Daniel Khashabi, Elad Segal, Tushar Khot, Dan Roth, and Jonathan Berant. Did aristotle use a laptop? a question answering benchmark with implicit reasoning strategies. Transactions of the Association for Computational Linguistics, 9:346–361, 2021. find
  14. Amelia Glaese, Nat McAleese, Maja Tr˛ebacz, John Aslanides, Vlad Firoiu, Timo Ewalds, Maribeth Rauh, Laura Weidinger, Martin Chadwick, Phoebe Thacker, et al. Improving alignment of dialogue agents via targeted human judgements. arXiv preprint arXiv:2209.14375, 2022.
  15. Hussein Hazimeh, Zhe Zhao, Aakanksha Chowdhery, Maheswaran Sathiamoorthy, Yihua Chen, Rahul Mazumder, Lichan Hong, and Ed H. Chi. Dselect-k: Differentiable selection in the mixture of experts with applications to multi-task learning. In Advances in Neural Information Processing Systems 34: Annual Co… find
  16. Dan Hendrycks, Collin Burns, Steven Basart, Andy Zou, Mantas Mazeika, Dawn Song, and Jacob Steinhardt. Measuring massive multitask language understanding. arXiv preprint arXiv:2009.03300, 2020.
  17. Or Honovich, Thomas Scialom, Omer Levy, and Timo Schick. Unnatural instructions: Tuning language models with (almost) no human labor. arXiv preprint arXiv:2212.09689, 2022.
  18. Srinivasan Iyer, Xi Victoria Lin, Ramakanth Pasunuru, Todor Mihaylov, Dániel Simig, Ping Yu, Kurt Shuster, Tianlu Wang, Qing Liu, Punit Singh Koura, et al. Opt-iml: Scaling language model instruction meta learning through the lens of generalization. arXiv preprint arXiv:2212.12017, 2022.
  19. Nitish Shirish Keskar, Bryan McCann, Caiming Xiong, and Richard Socher. Unifying question answering, text classification, and regression via span extraction. arXiv preprint arXiv:1904.09286, 2019. — [arXiv:1904.09286](https://arxiv.org/abs/1904.09286) find
  20. Daniel Khashabi, Sewon Min, Tushar Khot, Ashish Sabharwal, Oyvind Tafjord, Peter Clark, and Hannaneh Hajishirzi. Unifiedqa: Crossing format boundaries with a single qa system. arXiv preprint arXiv:2005.00700, 2020.
  21. Aran Komatsuzaki, Joan Puigcerver, James Lee-Thorp, Carlos Riquelme Ruiz, Basil Mustafa, Joshua Ainslie, Yi Tay, Mostafa Dehghani, and Neil Houlsby. Sparse upcycling: Training mixture-of-experts from dense checkpoints. arXiv preprint arXiv:2212.05055, 2022. find
  22. Sneha Kudugunta, Yanping Huang, Ankur Bapna, Maxim Krikun, Dmitry Lepikhin, Minh-Thang Luong, and Orhan Firat. Beyond distillation: Task-level mixture-of-experts for efficient inference. In Findings of the Association for Computational Linguistics: EMNLP 2021, pages 3577–3599, 2021. find
  23. Dmitry Lepikhin, HyoukJoong Lee, Yuanzhong Xu, Dehao Chen, Orhan Firat, Yanping Huang, Maxim Krikun, Noam Shazeer, and Zhifeng Chen. Gshard: Scaling giant models with conditional computation and automatic sharding. arXiv preprint arXiv:2006.16668, 2020. find
  24. Mike Lewis, Shruti Bhosale, Tim Dettmers, Naman Goyal, and Luke Zettlemoyer. BASE layers: Simplifying training of large, sparse models. In ICML. PMLR, 2021. find
  25. Bo Li, Yuanhan Zhang, Liangyu Chen, Jinghao Wang, Jingkang Yang, and Ziwei Liu. Otter: A multi-modal model with in-context instruction tuning. arXiv preprint arXiv:2305.03726, 2023.
  26. Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. Visual instruction tuning. arXiv preprint arXiv:2304.08485, 2023.
  27. Xiaodong Liu, Pengcheng He, Weizhu Chen, and Jianfeng Gao. Multi-task deep neural networks for natural language understanding. arXiv preprint arXiv:1901.11504, 2019.
  28. Shayne Longpre, Le Hou, Tu Vu, Albert Webson, Hyung Won Chung, Yi Tay, Denny Zhou, Quoc V Le, Barret Zoph, Jason Wei, et al. The flan collection: Designing data and methods for effective instruction tuning. In ICML, 2023.
  29. Yuxuan Lou, Fuzhao Xue, Zangwei Zheng, and Yang You. Cross-token modeling with conditional computation. arXiv preprint arXiv:2109.02008, 2021. — [arXiv:2109.02008](https://arxiv.org/abs/2109.02008) find
  30. Jiaqi Ma, Zhe Zhao, Xinyang Yi, Jilin Chen, Lichan Hong, and Ed H. Chi. Modeling task relationships in multi-task learning with multi-gate mixture-of-experts. In Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, KDD 2018, London, UK, August 19-23, 2018… find
  31. Bryan McCann, Nitish Shirish Keskar, Caiming Xiong, and Richard Socher. The natural language decathlon: Multitask learning as question answering. arXiv preprint arXiv:1806.08730, 2018.
  32. Shen-Yun Miao, Chao-Chun Liang, and Keh-Yih Su. A diverse corpus for evaluating and developing english math word problem solvers. In ACL, 2020. find
  33. Swaroop Mishra, Daniel Khashabi, Chitta Baral, and Hannaneh Hajishirzi. Cross-task generalization via natural language crowdsourcing instructions. arXiv preprint arXiv:2104.08773, 2021. find
  34. Niklas Muennighoff, Thomas Wang, Lintang Sutawika, Adam Roberts, Stella Biderman, Teven Le Scao, M Saiful Bari, Sheng Shen, Zheng-Xin Yong, Hailey Schoelkopf, et al. Crosslingual generalization through multitask finetuning. arXiv preprint arXiv:2211.01786, 2022.
  35. Niklas Muennighoff, Thomas Wang, Lintang Sutawika, Adam Roberts, Stella Biderman, Teven Le Scao, M Saiful Bari, Sheng Shen, Zheng-Xin Yong, Hailey Schoelkopf, et al. Crosslingual generalization through multitask finetuning. arXiv preprint arXiv:2211.01786, 2022.
  36. Basil Mustafa, Carlos Riquelme, Joan Puigcerver, Rodolphe Jenatton, and Neil Houlsby. Multimodal contrastive learning with limoe: the language-image mixture of experts. arXiv preprint arXiv:2206.02770, 2022. — [arXiv:2206.02770](https://arxiv.org/abs/2206.02770) find
  37. Reiichiro Nakano, Jacob Hilton, Suchir Balaji, Jeff Wu, Long Ouyang, Christina Kim, Christopher Hesse, Shantanu Jain, Vineet Kosaraju, William Saunders, et al. Webgpt: Browser-assisted question-answering with human feedback. arXiv preprint arXiv:2112.09332, 2021.
  38. Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al. Training language models to follow instructions with human feedback. In Advances in Neural Information Processing Systems, volume 35, pages 27730–2774…
  39. Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al. Training language models to follow instructions with human feedback. Advances in Neural Information Processing Systems, 35:27730–27744, 2022.
  40. Arkil Patel, Satwik Bhattamishra, and Navin Goyal. Are nlp models really able to solve simple math word problems? arXiv preprint arXiv:2103.07191, 2021.
  41. Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, and Peter J. Liu. Exploring the limits of transfer learning with a unified text-to-text transformer. J. Mach. Learn. Res., 21:140:1–140:67, 2020.
  42. Carlos Riquelme, Joan Puigcerver, Basil Mustafa, Maxim Neumann, Rodolphe Jenatton, André Susano Pinto, Daniel Keysers, and Neil Houlsby. Scaling vision with sparse mixture of experts. Advances in Neural Information Processing Systems, 2021. find
  43. Stephen Roller, Sainbayar Sukhbaatar, Arthur Szlam, and Jason Weston. Hash layers for large sparse models. In Advances in Neural Information Processing Systems 34: Annual Conference on Neural Information Processing Systems 2021, NeurIPS 2021, December 6-14, 2021, virtual, 2021. find
  44. Victor Sanh, Albert Webson, Colin Raffel, Stephen H Bach, Lintang Sutawika, Zaid Alyafeai, Antoine Chaffin, Arnaud Stiegler, Teven Le Scao, Arun Raja, et al. Multitask prompted training enables zero-shot task generalization. In ICLR, 2022.
  45. Noam Shazeer, Azalia Mirhoseini, Krzysztof Maziarz, Andy Davis, Quoc V. Le, Geoffrey E. Hinton, and Jeff Dean. Outrageously large neural networks: The sparsely-gated mixture-of-experts layer. In ICLR. OpenReview.net, 2017. find
  46. Sheng Shen, Zhewei Yao, Chunyuan Li, Trevor Darrell, Kurt Keutzer, and Yuxiong He. Scaling vision-language models with sparse mixture of experts. arXiv preprint arXiv:2303.07226, 2023. find
  47. Aarohi Srivastava, Abhinav Rastogi, Abhishek Rao, Abu Awal Md Shoeb, Abubakar Abid, Adam Fisch, Adam R Brown, Adam Santoro, Aditya Gupta, Adrià Garriga-Alonso, et al. Beyond the imitation game: Quantifying and extrapolating the capabilities of language models. arXiv preprint arXiv:2206.04615, 2022.
  48. Mirac Suzgun, Nathan Scales, Nathanael Schärli, Sebastian Gehrmann, Yi Tay, Hyung Won Chung, Aakanksha Chowdhery, Quoc V Le, Ed H Chi, Denny Zhou, et al. Challenging big-bench tasks and whether chain-of-thought can solve them. arXiv preprint arXiv:2210.09261, 2022.
  49. Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser, and Illia Polosukhin. Attention is all you need. In Isabelle Guyon, Ulrike von Luxburg, Samy Bengio, Hanna M. Wallach, Rob Fergus, S. V. N. Vishwanathan, and Roman Garnett, editors, Advances in Ne…
  50. Yizhong Wang, Yeganeh Kordi, Swaroop Mishra, Alisa Liu, Noah A Smith, Daniel Khashabi, and Hannaneh Hajishirzi. Self-instruct: Aligning language model with self generated instructions. arXiv preprint arXiv:2212.10560, 2022. find
  51. Yizhong Wang, Swaroop Mishra, Pegah Alipoormolabashi, Yeganeh Kordi, Amirreza Mirzaei, Anjana Arunkumar, Arjun Ashok, Arut Selvan Dhanasekaran, Atharva Naik, David Stap, et al. Super-naturalinstructions: Generalization via declarative instructions on 1600+ nlp tasks. In EMNLP, 2022. find
  52. Jason Wei, Maarten Bosma, Vincent Y Zhao, Kelvin Guu, Adams Wei Yu, Brian Lester, Nan Du, Andrew M Dai, and Quoc V Le. Finetuned language models are zero-shot learners. In ICLR, 2022. find
  53. Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Ed Chi, Quoc Le, and Denny Zhou. Chain of thought prompting elicits reasoning in large language models. In NeurIPS, 2022.
  54. Qinyuan Ye, Bill Yuchen Lin, and Xiang Ren. Crossfit: A few-shot learning challenge for cross-task generalization in nlp. arXiv preprint arXiv:2104.08835, 2021. find
  55. Yanqi Zhou, Tao Lei, Hanxiao Liu, Nan Du, Yanping Huang, Vincent Y Zhao, Andrew M Dai, Zhifeng Chen, Quoc V Le, and James Laudon. Mixture-of-experts with expert choice routing. In Advances in Neural Information Processing Systems, 2022. find
  56. Barret Zoph, Irwan Bello, Sameer Kumar, Nan Du, Yanping Huang, Jeff Dean, Noam Shazeer, and William Fedus. St-moe: Designing stable and transferable sparse expert models. arXiv preprint arXiv:2202.08906, 2022. find
  57. Simiao Zuo, Xiaodong Liu, Jian Jiao, Young Jin Kim, Hany Hassan, Ruofei Zhang, Tuo Zhao, and Jianfeng Gao. Taming sparsely activated transformer with stochastic experts. arXiv preprint arXiv:2110.04260, 2021. Appendix for “Mixture-of-Experts Meets Instruction Tuning: A Winning Combination for Large… — [arXiv:2110.04260](https://arxiv.org/abs/2110.04260) find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/shen-2023-mixture-experts-meets-instruction-tuning/