author
Applied Research Scientist
,
Brown University
Eliot Horowitz Assistant Professor, Computer Science Department
Stephen Bach is the Eliot Horowitz Assistant Professor in the Computer Science Department at Brown University. Previously, he was a visiting scholar at Google, and a postdoctoral scholar in the computer science department at Stanford University advised by Christopher Ré.
He received his Ph.D. in computer science from the University of Maryland, where he was advised by Lise Getoor. His research focuses on weakly supervised, zero-shot, and few-shot machine learning. The goal of his work is to create methods and systems that drive down the labor cost of AI. He was a core contributor to the Snorkel framework, which was recognized with a Best of VLDB 2018 award. He also co-led the team that developed the T0 family of large language models. The team was also one of the proposers of instruction tuning, which is the process of fine-tuning language models with supervised training to follow instructions. Instruction tuning is now a standard part of training large language models. Stephen is also an advisor to Snorkel AI.
The latest from Stephen
Tight Lower Bounds on Worst-Case Guarantees for Zero-Shot Learning with Attributes
This paper demonstrates a mathematical analysis of zero-shot learning with attributes, providing a tight lower bound on the worst-case error of the best map from attributes to classes and showing that this bound is predictive of how standard zero-shot methods behave in practice.
Research Paper
Tight Lower Bounds on Worst-Case Guarantees for Zero-Shot Learning with Attributes
This paper demonstrates a mathematical analysis of zero-shot learning with attributes, providing a tight lower bound on the worst-case error of the best map from attributes to classes and showing that this bound is predictive of how standard zero-shot methods behave in practice.
PromptSource: An Integrated Development Environment and Repository for Natural Language Prompts
PromptSource is a system that provides a templating language, an interface, and a set of guidelines to create, share, and use natural language prompts to train and query language models.
Research Paper
PromptSource: An Integrated Development Environment and Repository for Natural Language Prompts
PromptSource is a system that provides a templating language, an interface, and a set of guidelines to create, share, and use natural language prompts to train and query language models.
Blog
Better not bigger: How to get GPT-3 quality at 0.1% the cost
We created Data-centric Foundation Model Development to bridge the gaps between foundation models and enterprise AI. New Snorkel Flow capabilities (Foundation Model Fine-tuning, Warm Start, and Prompt Builder) give data science and machine learning teams the tools they need to effectively put foundation models (FMs) to use for performance-critical enterprise use cases. The need is clear: despite undeniable excitement about…
TAGLETS: A System for Automatic Semi-Supervised Learning with Auxiliary Data
This paper describes TAGLETS, a system built to study techniques for automatically exploiting all three types of data and creating high-quality, servable classifiers
Research Paper
TAGLETS: A System for Automatic Semi-Supervised Learning with Auxiliary Data
This paper describes TAGLETS, a system built to study techniques for automatically exploiting all three types of data and creating high-quality, servable classifiers
Multitask prompted training enables zero-shot task generalization
This paper showcases how using a data-centric approach to generate high-quality training data at massive scale to improve the zero-shot abilities of that model.
Research Paper
Multitask prompted training enables zero-shot task generalization
This paper showcases how using a data-centric approach to generate high-quality training data at massive scale to improve the zero-shot abilities of that model.
Learning from Multiple Noisy Partial Labelers
This work enables users to create partial labelers that output subsets of possible class labels would greatly expand the expressivity of programmatic weak supervision.
Research Paper
Learning from Multiple Noisy Partial Labelers
This work enables users to create partial labelers that output subsets of possible class labels would greatly expand the expressivity of programmatic weak supervision.
Semi-Supervised Aggregation of Dependent Weak Supervision Sources with Performance Guarantees
This work shows a rigorous technique for efficiently selecting small subsets of the labelers so that a majority vote from such subsets has a provably low error rate.
Research Paper
Semi-Supervised Aggregation of Dependent Weak Supervision Sources with Performance Guarantees
This work shows a rigorous technique for efficiently selecting small subsets of the labelers so that a majority vote from such subsets has a provably low error rate.
Extended Few-Shot Learning: Exploiting Existing Resources for Novel Tasks
In many practical few-shot learning problems, even though labeled examples are scarce, there are abundant auxiliary datasets that potentially contain useful information. We propose the problem of extended few-shot learning to study these scenarios. We then introduce a framework to address the challenges of efficiently selecting and effectively using auxiliary data in few-shot image classification. Given a large auxiliary dataset and a notion of semantic similarity among classes, we automatically select pseudo shots, which are labeled examples from other classes related to the target task. We show that naive approaches, such as (1) modeling these additional examples the same as...
Research Paper
Extended Few-Shot Learning: Exploiting Existing Resources for Novel Tasks
In many practical few-shot learning problems, even though labeled examples are scarce, there are abundant auxiliary datasets that potentially contain useful information. We propose the problem of extended few-shot learning to study these scenarios. We then introduce a framework to address the challenges of efficiently selecting and effectively using auxiliary data in few-shot image classification. Given a large auxiliary dataset…
What will it take to generate fairness-preserving explanations?
In situations where explanations of black-box models may be useful, the fairness of the blackbox is also often a relevant concern. However, the link between the fairness of the black-box model and the behavior of explanations for the black-box is unclear. We focus on explanations applied to tabular datasets, suggesting that explanations do not necessarily preserve the fairness properties of the black-box algorithm. In other words, explanation algorithms can ignore or obscure critical relevant properties, creating incorrect or misleading explanations. More broadly, we propose future research directions for evaluating and generating explanations such that they are informative and relevant from...
Research Paper
What will it take to generate fairness-preserving explanations?
In situations where explanations of black-box models may be useful, the fairness of the blackbox is also often a relevant concern. However, the link between the fairness of the black-box model and the behavior of explanations for the black-box is unclear. We focus on explanations applied to tabular datasets, suggesting that explanations do not necessarily preserve the fairness properties of…
