Publications (34)
Learning to Make Analogies by Contrasting Abstract Relational Structure
Felix Hill, Adam Santoro, David G. T. Barrett +2
Analogical reasoning has been a principal focus of various waves of AI research. Analogy is particularly challenging for machines because it requires relational structures to be re…
A simple neural network module for relational reasoning
Adam Santoro, David Raposo, David G. T. Barrett +4
Relational reasoning is a central component of generally intelligent behavior, but has proven difficult for neural networks to learn. In this paper we describe how to use Relation…
Relational inductive biases, deep learning, and graph networks
Peter W. Battaglia, Jessica B. Hamrick, Victor Bapst +24
Artificial intelligence (AI) has undergone a renaissance recently, making major progress in key domains such as vision, language, control, and decision-making. This has been due, i…
Intra-agent speech permits zero-shot task acquisition
Chen Yan, Federico Carnevale, Petko Georgiev +7
Human language learners are exposed to a trickle of informative, context-sensitive language, but a flood of raw sensory data. Through both social language use and internal processe…
Imitating Interactive Intelligence
Josh Abramson, Arun Ahuja, Iain Barr +26
A common vision from science fiction is that robots will one day inhabit our physical spaces, sense the world as we do, assist our physical labours, and communicate with us through…
Symbolic Behaviour in Artificial Intelligence
Adam Santoro, Andrew Lampinen, Kory Mathewson +2
The ability to use symbols is the pinnacle of human intelligence, but has yet to be fully replicated in machines. Here we argue that the path towards symbolically fluent artificial…
Automated curricula through setter-solver interactions
Sebastien Racaniere, Andrew K. Lampinen, Adam Santoro +3
Reinforcement learning algorithms use correlations between policies and rewards to improve agent performance. But in dynamic or sparsely rewarding environments these correlations a…
Hyperbolic Attention Networks
Caglar Gulcehre, Misha Denil, Mateusz Malinowski +8
We introduce hyperbolic attention networks to endow neural networks with enough capacity to match the complexity of data with hierarchical and power-law structure. A few recent app…
Environmental drivers of systematicity and generalization in a situated agent
Felix Hill, Andrew Lampinen, Rosalia Schneider +4
The question of whether deep neural networks are good at generalising beyond their immediate training experience is of critical importance for learning-based approaches to AI. Here…
A data-driven approach for learning to control computers
Peter C Humphreys, David Raposo, Toby Pohlen +8
It would be useful for machines to use computers as humans do so that they can aid us in everyday tasks. This is a setting in which there is also the potential to leverage large-sc…
Assessing the Scalability of Biologically-Motivated Deep Learning Algorithms and Architectures
Sergey Bartunov, Adam Santoro, Blake A. Richards +3
The backpropagation of error algorithm (BP) is impossible to implement in a real brain. The recent success of deep networks in machine learning and AI, however, has inspired propos…
Creating Multimodal Interactive Agents with Imitation and Self-Supervised Learning
DeepMind Interactive Agents Team, Josh Abramson, Arun Ahuja +22
A common vision from science fiction is that robots will one day inhabit our physical spaces, sense the world as we do, assist our physical labours, and communicate with us through…
Relational Deep Reinforcement Learning
Vinicius Zambaldi, David Raposo, Adam Santoro +13
We introduce an approach for deep reinforcement learning (RL) that improves upon the efficiency, generalization capacity, and interpretability of conventional approaches through st…
Learning Visual Question Answering by Bootstrapping Hard Attention
Mateusz Malinowski, Carl Doersch, Adam Santoro +1
Attention mechanisms in biological perception are thought to select subsets of perceptual information for more sophisticated processing which would be prohibitive to perform on all…
Unsupervised Predictive Memory in a Goal-Directed Agent
Greg Wayne, Chia-Chun Hung, David Amos +21
Animals execute goal-directed behaviours despite the limited range and scope of their sensors. To cope, they explore environments and store memories maintaining estimates of import…
Improving Multimodal Interactive Agents with Reinforcement Learning from Human Feedback
Josh Abramson, Arun Ahuja, Federico Carnevale +16
An important goal in artificial intelligence is to create agents that can both interact naturally with humans and learn from their feedback. Here we demonstrate how to use reinforc…
Rapid Task-Solving in Novel Environments
Sam Ritter, Ryan Faulkner, Laurent Sartran +3
We propose the challenge of rapid task-solving in novel environments (RTS), wherein an agent must solve a series of tasks as rapidly as possible in an unfamiliar environment. An ef…
Relational recurrent neural networks
Adam Santoro, Ryan Faulkner, David Raposo +7
Memory-based neural networks model temporal data by leveraging an ability to remember information for long periods. It is unclear, however, whether they also have an ability to per…
One-shot Learning with Memory-Augmented Neural Networks
Adam Santoro, Sergey Bartunov, Matthew Botvinick +2
Despite recent breakthroughs in the applications of deep neural networks, one setting that presents a persistent challenge is that of "one-shot learning." Traditional gradient-base…
Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities
Gheorghe Comanici, Eric Bieber, Mike Schaekermann +3431
In this report, we introduce the Gemini 2.X model family: Gemini 2.5 Pro and Gemini 2.5 Flash, as well as our earlier Gemini 2.0 Flash and Flash-Lite models. Gemini 2.5 Pro is our…
Tracing the Representation Geometry of Language Models from Pretraining to Post-training
Melody Zixuan Li, Kumar Krishna Agrawal, Arna Ghosh +4
Standard training metrics like loss fail to explain the emergence of complex capabilities in large language models. We take a spectral approach to investigate the geometry of learn…
Is coding a relevant metaphor for building AI? A commentary on "Is coding a relevant metaphor for the brain?", by Romain Brette
Adam Santoro, Felix Hill, David Barrett +3
Brette contends that the neural coding metaphor is an invalid basis for theories of what the brain does. Here, we argue that it is an insufficient guide for building an artificial…
Cognitive Psychology for Deep Neural Networks: A Shape Bias Case Study
Samuel Ritter, David G. T. Barrett, Adam Santoro +1
Deep neural networks (DNNs) have achieved unprecedented performance on a wide range of complex tasks, rapidly outpacing our understanding of the nature of their solutions. This has…
Tell me why! Explanations support learning relational and causal structure
Andrew K. Lampinen, Nicholas A. Roy, Ishita Dasgupta +8
Inferring the abstract relational and causal structure of the world is a major challenge for reinforcement-learning (RL) agents. For humans, language--particularly in the form of e…
An investigation of model-free planning
Arthur Guez, Mehdi Mirza, Karol Gregor +10
The field of reinforcement learning (RL) is facing increasingly challenging domains with combinatorial complexity. For an RL agent to address these challenges, it is essential that…
Discovering objects and their relations from entangled scene representations
David Raposo, Adam Santoro, David Barrett +3
Our world can be succinctly and compactly described as structured scenes of objects and relations. A typical room, for example, contains salient objects such as tables, chairs and…
Data Distributional Properties Drive Emergent In-Context Learning in Transformers
Stephanie C. Y. Chan, Adam Santoro, Andrew K. Lampinen +5
Large transformer-based models are able to perform in-context few-shot learning, without being explicitly trained for it. This observation raises the question: what aspects of the…
Evaluating Multimodal Interactive Agents
Josh Abramson, Arun Ahuja, Federico Carnevale +12
Creating agents that can interact naturally with humans is a common goal in artificial intelligence (AI) research. However, evaluating these interactions is challenging: collecting…
Beyond the Imitation Game: Quantifying and extrapolating the capabilities of language models
Aarohi Srivastava, Abhinav Rastogi, Abhishek Rao +448
Language models demonstrate both quantitative improvement and new qualitative capabilities with increasing scale. Despite their potentially transformative impact, these new capabil…
Attention over learned object embeddings enables complex visual reasoning
David Ding, Felix Hill, Adam Santoro +2
Neural networks have achieved success in a wide array of perceptual tasks but often fail at tasks involving both perception and higher-level reasoning. On these more challenging ta…
Measuring abstract reasoning in neural networks
David G. T. Barrett, Felix Hill, Adam Santoro +2
Whether neural networks can learn abstract reasoning or whether they merely rely on superficial statistics is a topic of recent debate. Here, we propose a dataset and challenge des…
Generative Temporal Models with Memory
Mevlana Gemici, Chia-Chun Hung, Adam Santoro +5
We consider the general problem of modeling temporal data with long-range dependencies, wherein new observations are fully or partially predictable based on temporally-distant, pas…
Mixture-of-Depths: Dynamically allocating compute in transformer-based language models
David Raposo, Sam Ritter, Blake Richards +3
Transformer-based language models spread FLOPs uniformly across input sequences. In this work we demonstrate that transformers can instead learn to dynamically allocate FLOPs (or c…
Synthetic Returns for Long-Term Credit Assignment
David Raposo, Sam Ritter, Adam Santoro +5
Since the earliest days of reinforcement learning, the workhorse method for assigning credit to actions over time has been temporal-difference (TD) learning, which propagates credi…