Last Updated: September 2023.
An up to date list of all publications can be found on my Google Scholar profile.

Filter: , , , , , , , , , , , , , ,


Learning to (Learn at Test Time): RNNs with Expressive Hidden States

Yu Sun, Xinhao Li, Karan Dalal, Jiarui Xu, Arjun Vikram, Genghan Zhang, Y. Dubois, Xinlei Chen, Xiaolong Wang, Sanmi Koyejo, Tatsunori Hashimoto, Carlos Guestrin


TLDR: A new language model layer that is more expressive than RNN but more efficient than attention.


Length-Controlled AlpacaEval: A Simple Way to Debias Automatic Evaluators

Y. Dubois, B. Galambosi, P. Liang, T. Hashimoto

COLM 2024

TLDR: We decrease the bias of AlpacaEval for longer outputs using regression analysis.


HELM Instruct: A Multidimensional Instruction Following Evaluation Framework with Absolute Ratings

Y. Zhang, Y. Mai, J. Somerville Roberts, R. Bommasani, Y. Dubois, P. Liang


TLDR: Multidimensional evaluation of instruction following LLM with absolute scores.



Identifying the Risks of LM Agents with an LM-Emulated Sandbox

Y. Ruan*, H. Dong*, A. Wang, S. Pitis, Y. Zhou, J. Ba, Y. Dubois, C. J. Maddison, T. Hashimoto

ICLR 2024 Spotlight Presentation πŸŽ‰

TLDR: An LM-based emulation framework for identifying the risks of LM agents at scale.

, , ,

AlpacaEval: An Automatic Evaluator of Instruction-following Models

X. Li*, T. Zhang*, Y. Dubois*, R. Taori*, I. Gulrajani, C. Guestrin, P. Liang, T. Hashimoto


TLDR: A validated automatic evaluator for instruction-following language models. High-quality, cheap, and fast.

, , ,

AlpacaFarm: A Simulation Framework for Methods that Learn from Human Feedback

Y. Dubois*, X. Li*, R. Taori*, T. Zhang*, I. Gulrajani, J. Ba, C. Guestrin, P. Liang, T. Hashimoto

NeurIPS 2023 Spotlight Presentation πŸŽ‰

TLDR: AlpacaFarm replicates the RLHF process at a fraction of the time (<24h) and cost ($<200), enabling the research community to advance instruction following research.

, , ,

Stanford alpaca: An instruction-following llama model

R. Taori*, I. Gulrajani*, T. Zhang*, Y. Dubois*, X. Li*, C. Guestrin, P. Liang, T. Hashimoto


TLDR: We introduce Alpaca 7B, a instruction-following fine-tuned LLaMA model. On our preliminary evaluation, Alpaca behaves qualitatively similarly to OpenAI’s text-davinci-003, while being surprisingly small and easy/cheap to reproduce (<600$).

, ,

Evaluating Self-Supervised Learning via Risk Decomposition

Y. Dubois, T. Hashimoto, P. Liang

ICML 2023 Oral Presentation πŸŽ‰

TLDR: We derive a risk decomposition for self-supervised learning and use it to evaluate 169 pretrained models.

, , ,


Is a Caption Worth a Thousand Images? A Controlled Study for Representation Learning

S. Santurkar, Y. Dubois, R. Taori, P. Liang, T. Hashimoto

ICLR 2022

TLDR: Our work performs a systematic investigation into whether additional language supervision (in CLIP) helps models learn more transferrable representations.

, , ,

Learning Instance-Specific Data Augmentations

N. Miao, E. Mathieu, Y. Dubois, T. Rainforth, Y. W. Teh, A. Foster, H. Kim

ICML 2023

TLDR: We introduce a method for automatically learning input-specific augmentations from data.


Improving Self-Supervised Learning by Characterizing Idealized Representations

Y. Dubois, T. Hashimoto, S. Ermon, P. Liang

NeurIPS 2022

TLDR: We characterize idealized self-supervised representations, which leads to actionable insights for improving SSL algorithms.

, , ,

Optimal Representations for Covariate Shifts

Y Ruan*, Y. Dubois*, C. J. Maddison

ICLR 2021

TLDR: We give a simple variational objective whose optima are exactly the set of representations that are robust under covariate shift.

, , , , ,


Lossy Compression for Lossless Prediction

Y. Dubois, B. Bloem-Reddy, K. Ullrich, C. J. Maddison

NeurIPS 2021 Spotlight Presentation πŸŽ‰

TLDR: We formalize compression with respect to ML algorithms rather than human perception.

, , , , ,


Learning Optimal Representations with the Decodable Information Bottleneck

Y. Dubois, D. Kiela, D. J. Schwab, R. Vedantam

NeurIPS 2020 Spotlight Presentation πŸŽ‰

TLDR: We characterize and approximate optimal representations for supervised learning.

, ,

the Neural Process Family

Y. Dubois*, J. Gordon*, A. Foong*


TLDR: A simple and unifying explanation of the neural process family, which are a collection of models that meta-learn a distribution over predictors.

, , , ,

Meta-Learning Stationary Stochastic Process Prediction with Convolutional Neural Processes

A. Y. K. Foong*, W. P. Bruinsma*, J. Gordon*, Y. Dubois, J. Requeima, R. E. Turner

NeurIPS 2020

TLDR: We propose a translation equivariant (latent) neural process.

, , , ,

Convolutional Conditional Neural Processes

J. Gordon*, W. P. Bruinsma*, A. Y. K. Foong, J. Requeima,Y. Dubois, R. E. Turner

ICLR 2020 Oral Presentation πŸŽ‰

TLDR: We propose a translation equivariant conditional neural process.

, , , ,

Location Attention for Extrapolation to Longer Sequences

Y. Dubois, Gautier Dagan, Dieuwke Hupkes, Elia Bruni

ACL 2020

TLDR: We propose an attention that improves extrapolation capacity of neural NLP models.