Blue Hours Seattle. 2022
Google Scholar / Github / Twitter / LinkedIn / Blog
Yao Fu 符尧. yao.fu@ed.ac.uk
I am a Ph.D. student at the University of Edinburgh (2020-) with professor Mirella Lapata.
I finished my M.S. at Columbia University (2018-2020) with professor John Cunningham and my B.S. at Peking University (2013-2018) with professor Yansong Feng.
Before Ph.D., I spent great time visiting professor Alexander Rush at Cornell Tech (2019-2020).
I study large-scale generative models for human language.
My research objective is to make large language models the next generation computational platforms and build a language model based application ecosystem together with the community. I am broadly interested in the following topics:
- Science driven scaling: to pursue scientific principles bebind scaling and use them to guide next-generation model development, where the subareas include data engineering, long context, efficiency, and science of language models
- Reasoning as agents: to study how deployed language models reason over complex problems and environment, where the subareas include coding, multi-agent games, chain-of-thought, and learning from feedback
I am expected to graduate in Dec 2023 and will be joining Google DeepMind in Mountain View as a full-time research scientist.
AGI has yet to come, so keep running.
Experiences
Academia
- 2020 - 2023. Ph.D. at University of Edinburgh. Large Language Models
- 2018 - 2020. M.S. at Columbia University. Deep Generative Models
- 2013 - 2018. B.S. at Peking University. Language Generation
Industry
- Jun - Sep 2023. MIT-IBM Waston AI Lab. Research Intern on Training Large Language Models
- Jul - Dec 2022. Allen Institute for AI. Research Intern on Language Model Reasoning
- Jan - Oct 2020. Alibaba DAMO Academy. Research Intern on Latent Variable Models
- May - Sep 2019. Tencent AI Lab. Research Intern on Structured Prediction
- Jan - Aug 2018. Bytedance AI Lab. Research Intern on Language Generation
Featured Research
- [Arxiv 2023] Improving Language Model Negotiation with Self-Play and In-Context Learning from AI Feedback [code][paper]
- Yao Fu, Hao Peng, Tushar Khot, and Mirella Lapata
- Two language models negotiate with each other and continuously improve their negotiation strategies by multi-round game playing and iterative in-context learning from AI feedback.
- [ICML 2023] Oral. Specializing Smaller Language Models towards Multi-Step Reasoning. [paper][Code]
- Yao Fu, Hao Peng, Litu Ou, Ashish Sabharwal, and Tushar Khot
- Trading language model’s generic ability for specialized math chain-of-thought ability.
- [ICLR 2023] Complexity-Based Prompting for Multi-Step Reasoning. [paper][code]
- Yao Fu, Hao Peng, Ashish Sabharwal, Peter Clark and Tushar Khot
- State-of-the-art reasoning performance on math word problems by prompting GPT3 with instances of complex reasoning chains.
- [ICML Deployable GenAI 2023] Chain-of-thougth Hub: Measuring LLMs’ Reasoning Performance [GitHub][paper]
- Yao Fu, Litu Ou, Mingyu Chen and Yuhao Wan
- Benchmarking large language models’ complex reasoning performance with chain-of-thought prompting
Featured Blog Posts
- [Dec 2022] How does GPT Obtain its Ability? Tracing Emergent Abilities of Language Models to their Sources [notion]
- Yao Fu, Hao Peng and Tushar Khot
- Analysing sources of emergent abilities of Large Language Models from first principle.
- Hacker News top 3 trending.
- [May 2023]. Towards Complex Reasoning: the Polaris of Large Language Models [notion]
- Yao Fu
- A roadmap towards building language models of strong reasoning capabilties. Covers the full development stages: pretraining, continue training, supervised finetuning, reinforcemeng learning, chain-of-thought prompting, and evaluation.
- [Jun 2023]. A Stage Review of Instruction Tuning [notion]
- Yao Fu
- A review of the development of LLaMA-based models after the release of ChatGPT and discusses the next challenges of Instruction Tuning.