Search Results for author:
Found 27 papers, 17 papers with code
Clio: Privacy-Preserving Insights into Real-World AI Use
How are AI assistants being used in the real world?
Sycophancy to Subterfuge: Investigating Reward-Tampering in Large Language Models
We construct a curriculum of increasingly sophisticated gameable environments and find that training on early-curriculum environments leads to more specification gaming on remaining environments.
Collective Constitutional AI: Aligning a Language Model with Public Input
There is growing consensus that language model (LM) developers should not be the sole deciders of LM behavior, creating a need for methods that enable the broader public to collectively shape the behavior of LM systems that affect them.
Bayesian Preference Elicitation with Language Models
We introduce OPEN (Optimal Preference Elicitation with Natural language) a framework that uses BOED to guide the choice of informative questions and an LM to extract features and translate abstract BOED queries into natural language questions.
Evaluating and Mitigating Discrimination in Language Model Decisions
We present a method for proactively evaluating the potential discriminatory impact of LMs in a wide range of use cases, including hypothetical use cases where they have not yet been deployed.
Social Contract AI: Aligning AI Assistants with Implicit Group Norms
We explore the idea of aligning an AI assistant by inverting a model of users' (unknown) preferences from observed interactions.
Codebook Features: Sparse and Discrete Interpretability for Neural Networks
In this setting, our approach overcomes the superposition problem by assigning states to distinct codes, and we find that we can make the neural network behave as if it is in a different state by activating the code for that state.
Eliciting Human Preferences with Language Models
Language models (LMs) can be directed to perform target tasks by using labeled examples or natural language prompts.
Studying Large Language Model Generalization with Influence Functions
When trying to gain better visibility into a machine learning model in order to understand and mitigate the associated risks, a potentially valuable source of evidence is: which training examples most contribute to a given behavior?
Towards Measuring the Representation of Subjective Global Opinions in Language Models
We first build a dataset, GlobalOpinionQA, comprised of questions and answers from cross-national surveys designed to capture diverse opinions on global issues across different countries.
Operationalising the Definition of General Purpose AI Systems: Assessing Four Approaches
The European Union's Artificial Intelligence (AI) Act is set to be a landmark legal instrument for regulating AI technology.
BenchMD: A Benchmark for Unified Learning on Medical Images and Sensors
Finally, we evaluate performance on out-of-distribution data collected at different hospitals than the training data, representing naturally-occurring distribution shifts that frequently degrade the performance of medical AI models.
Multispectral Contrastive Learning with Viewmaker Networks
Contrastive learning methods have been applied to a range of domains and modalities by training models to identify similar "views" of data points.
Task Ambiguity in Humans and Language Models
We investigate how both humans and models behave in the face of such task ambiguity by proposing AmbiBench, a new benchmark of six ambiguously-specified classification tasks.
Active Learning Helps Pretrained Models Learn the Intended Task
Models can fail in unpredictable ways during deployment due to task ambiguity, when multiple behaviors are consistent with the provided training data.
Oolong: Investigating What Makes Transfer Learning Hard with Controlled Studies
When we transfer a pretrained language model to a new language, there are many axes of variation that change at once.
Tradeoffs Between Contrastive and Supervised Learning: An Empirical Study
Contrastive learning has made considerable progress in computer vision, outperforming supervised pretraining on a range of downstream datasets.
DABS: A Domain-Agnostic Benchmark for Self-Supervised Learning
Self-supervised learning algorithms, including BERT and SimCLR, have enabled significant strides in fields like natural language processing, computer vision, and speech processing.
Ranked #1 on
Self-Supervised Learning
on DABS
Pretrained models are active learners
An important barrier to the safe deployment of machine learning systems is the risk of \emph{task ambiguity}, where multiple behaviors are consistent with the provided examples.
C5T5: Controllable Generation of Organic Molecules with Transformers
Methods for designing organic materials with desired properties have high potential impact across fields such as medicine, renewable energy, petrochemical engineering, and agriculture.
On the Opportunities and Risks of Foundation Models
AI is undergoing a paradigm shift with the rise of models (e. g., BERT, DALL-E, GPT-3) that are trained on broad data at scale and are adaptable to a wide range of downstream tasks.
Understanding the Capabilities, Limitations, and Societal Impact of Large Language Models
On October 14th, 2020, researchers from OpenAI, the Stanford Institute for Human-Centered Artificial Intelligence, and other universities convened to discuss open research questions surrounding GPT-3, the largest publicly-disclosed dense language model at the time.
Language Through a Prism: A Spectral Approach for Multiscale Language Representations
Language exhibits structure at different scales, ranging from subwords to words, sentences, paragraphs, and documents.
Viewmaker Networks: Learning Views for Unsupervised Representation Learning
However, designing these views requires considerable trial and error by human experts, hindering widespread adoption of unsupervised representation learning methods across domains and modalities.
Investigating Transferability in Pretrained Language Models
How does language model pretraining help transfer learning?
Being Optimistic to Be Conservative: Quickly Learning a CVaR Policy
While maximizing expected return is the goal in most reinforcement learning approaches, risk-sensitive objectives such as conditional value at risk (CVaR) are more suitable for many high-stakes applications.
Recursive Routing Networks: Learning to Compose Modules for Language Understanding
The model jointly optimizes the parameters of the functions and the meta-learner{'}s policy for routing inputs through those functions.
