Search Results for author:
Found 28 papers, 13 papers with code
Holistic Safety and Responsibility Evaluations of Advanced AI Models
In the development of Google DeepMind's advanced AI models, we innovated on and applied a broad set of approaches to safety evaluation.
Evaluating Frontier Models for Dangerous Capabilities
To understand the risks posed by a new AI system, we must understand what it can and cannot do.
Challenges with unsupervised LLM knowledge discovery
We show that existing unsupervised methods on large language model (LLM) activations do not discover knowledge -- instead they seem to discover whatever feature of the activations is most prominent.
Model evaluation for extreme risks
Current approaches to building general-purpose AI systems tend to produce systems with both beneficial and harmful capabilities.
Prediction-Oriented Bayesian Active Learning
Information-theoretic approaches to active learning have traditionally focused on maximising the information gathered about the model parameters, most commonly by optimising the BALD score.
Semantic Uncertainty: Linguistic Invariances for Uncertainty Estimation in Natural Language Generation
We introduce a method to measure uncertainty in large language models.
Tracr: Compiled Transformers as a Laboratory for Interpretability
Additionally, the known structure of Tracr-compiled models can serve as ground-truth for evaluating interpretability methods.
CLAM: Selective Clarification for Ambiguous Questions with Generative Language Models
Users often ask dialogue systems ambiguous questions that require clarification.
Understanding Approximation for Bayesian Inference in Neural Networks
To assess a model's ability to incorporate different parts of the Bayesian framework we can identify desirable characteristic behaviours of Bayesian reasoning and pick decision-problems that make heavy use of those behaviours.
Do Bayesian Neural Networks Need To Be Fully Stochastic?
We investigate the benefit of treating all the parameters in a Bayesian neural network stochastically and find compelling theoretical and empirical evidence that this standard construction may be unnecessary.
Discovering Agents
Causal models of agents have been used to analyse the safety aspects of machine learning systems.
Prioritized Training on Points that are Learnable, Worth Learning, and Not Yet Learnt
But most computation and time is wasted on redundant and noisy points that are already learnt or not learnable.
Path-Specific Objectives for Safer Agent Incentives
We then train agents to maximize the causal effect of actions on the expected return which is not mediated by the delicate parts of state, using Causal Influence Diagram analysis.
Prospect Pruning: Finding Trainable Weights at Initialization using Meta-Gradients
Pruning neural networks at initialization would enable us to find sparse models that retain the accuracy of the original network while consuming fewer computational resources for training and inference.
Active Surrogate Estimators: An Active Learning Approach to Label-Efficient Model Evaluation
We propose Active Surrogate Estimators (ASEs), a new method for label-efficient model evaluation.
Prioritized training on points that are learnable, worth learning, and not yet learned (workshop version)
We introduce Goldilocks Selection, a technique for faster model training which selects a sequence of training points that are "just right".
Stochastic Batch Acquisition: A Simple Baseline for Deep Active Learning
We examine a simple stochastic strategy for adapting well-known single-point acquisition functions to allow batch active learning.
Uncertainty Baselines: Benchmarks for Uncertainty & Robustness in Deep Learning
In this paper we introduce Uncertainty Baselines: high-quality implementations of standard and state-of-the-art deep learning methods on a variety of tasks.
Active Testing: Sample-Efficient Model Evaluation
While approaches like active learning reduce the number of labels needed for model training, existing literature largely ignores the cost of labeling test data, typically unrealistically assuming large test sets for model evaluation.
On Statistical Bias In Active Learning: How and When To Fix It
Active learning is a powerful tool when labelling data is expensive, but it introduces a bias because the training data no longer follows the population distribution.
Single Shot Structured Pruning Before Training
We introduce a method to speed up training by 2x and inference by 3x in deep neural networks using structured pruning applied before training.
Liberty or Depth: Deep Bayesian Neural Nets Do Not Need Complex Weight Posterior Approximations
We challenge the longstanding assumption that the mean-field approximation for variational inference in Bayesian neural networks is severely restrictive, and show this is not the case in deep networks.
A Systematic Comparison of Bayesian Deep Learning Robustness in Diabetic Retinopathy Tasks
From our comparison we conclude that some current techniques which solve benchmarks such as UCI `overfit' their uncertainty to the dataset---when evaluated on our benchmark these underperform in comparison to simpler baselines.
Radial Bayesian Neural Networks: Beyond Discrete Support In Large-Scale Bayesian Deep Learning
The Radial BNN is motivated by avoiding a sampling problem in 'mean-field' variational inference (MFVI) caused by the so-called 'soap-bubble' pathology of multivariate Gaussians.
A Unifying Bayesian View of Continual Learning
From a Bayesian perspective, continual learning seems straightforward: Given the model posterior one would simply use this as the prior for the next task.
Differentially Private Continual Learning
Catastrophic forgetting can be a significant problem for institutions that must delete historic data for privacy reasons.
Towards Robust Evaluations of Continual Learning
Experiments used in current continual learning research do not faithfully assess fundamental challenges of learning continually.
The Malicious Use of Artificial Intelligence: Forecasting, Prevention, and Mitigation
This report surveys the landscape of potential security threats from malicious uses of AI, and proposes ways to better forecast, prevent, and mitigate these threats.
