AIConfPaper
← Search

Andrea Banino

8 accepted papers

2024

Transformers need glasses! Information over-squashing in language tasks

Federico Barbero, Andrea Banino, Steven Kapturowski, Dharshan Kumaran, João Guilherme Madeira Araújo, Alex Vitvitskyi +2
NeurIPS 2024poster

We study how information propagates in decoder-only Transformers, which are the architectural foundation of most existing frontier large language models (LLMs). We rely on a theoretical signal propagation analysis---specifically, we analyse the representations of the last token in the final layer of…

Cited by 21Source ↗PDF ↗Scholar ↗
2022

CoBERL: Contrastive BERT for Reinforcement Learning

Andrea Banino, Adria Puigdomenech Badia, Jacob C Walker, Tim Scholtes, Jovana Mitrovic, Charles Blundell
ICLR 2022spotlight

Many reinforcement learning (RL) agents require a large amount of experience to solve tasks. We propose Contrastive BERT for RL (COBERL), an agent that combines a new contrastive loss and a hybrid LSTM-transformer architecture to tackle the challenge of improving data efficiency. COBERL enables effi…

Cited by 46Source ↗PDF ↗code ↗Scholar ↗
2022

Retrieval-Augmented Reinforcement Learning

Anirudh Goyal, Abram Friesen, Andrea Banino, Theophane Weber, Nan Rosemary Ke, Adrià Puigdomènech Badia +9
ICML 2022spotlight

Most deep reinforcement learning (RL) algorithms distill experience into parametric behavior policies or value functions via gradient updates. While effective, this approach has several disadvantages: (1) it is computationally expensive, (2) it can take many updates to integrate experiences into the…

Cited by 59Source ↗PDF ↗Scholar ↗
2022

Semantic Exploration from Language Abstractions and Pretrained Representations

Allison Tam, Neil Charles Rabinowitz, Andrew Kyle Lampinen, Nicholas Andrew Roy, Stephanie C.Y. Chan, DJ Strouse +3
NeurIPS 2022accept

Effective exploration is a challenge in reinforcement learning (RL). Novelty-based exploration methods can suffer in high-dimensional state spaces, such as continuous partially-observable 3D environments. We address this challenge by defining novelty using semantically meaningful state abstractions,…

Cited by 72Source ↗PDF ↗Scholar ↗
2022

The CLRS Algorithmic Reasoning Benchmark

Petar Veličković, Adrià Puigdomènech Badia, David Budden, Razvan Pascanu, Andrea Banino, Misha Dashevskiy +2
ICML 2022spotlight

Learning representations of algorithms is an emerging area of machine learning, seeking to bridge concepts from neural networks with classical algorithms. Several important works have investigated whether neural networks can effectively reason like algorithms, typically by learning to execute them.…

Cited by 119Source ↗PDF ↗code ↗Scholar ↗
2021

Towards mental time travel: a hierarchical memory for reinforcement learning agents

Andrew Kyle Lampinen, Stephanie C.Y. Chan, Andrea Banino, Felix Hill
NeurIPS 2021poster

Reinforcement learning agents often forget details of the past, especially after delays or distractor tasks. Agents with common memory architectures struggle to recall and integrate across multiple timesteps of a past event, or even to recall the details of a single timestep that is followed by dist…

Cited by 59Source ↗PDF ↗Scholar ↗
2020

MEMO: A Deep Network for Flexible Combination of Episodic Memories

Andrea Banino, Adrià Puigdomènech Badia, Raphael Köster, Martin J. Chadwick, Vinicius Zambaldi, Demis Hassabis +4
ICLR 2020poster

Recent research developing neural network architectures with external memory have often used the benchmark bAbI question and answering dataset which provides a challenging number of tasks requiring reasoning. Here we employed a classic associative inference task from the human neuroscience literatur…

Cited by 0Source ↗Scholar ↗
2017

Learning to Navigate in Complex Environments

Piotr Mirowski, Razvan Pascanu, Fabio Viola, Hubert Soyer, Andy Ballard, Andrea Banino +6
ICLR 2017poster

Learning to navigate in complex environments with dynamic elements is an important milestone in developing AI agents. In this work we formulate the navigation question as a reinforcement learning problem and show that data efficiency and task performance can be dramatically improved by relying on ad…

Cited by 1075Source ↗code ↗Scholar ↗