We propose a novel multimodal video benchmark - the Perception Test - to...
The unabated mystique of large-scale neural networks, such as the CLIP d...
We investigate the optimal model size and number of tokens for training ...
Most deep reinforcement learning (RL) algorithms distill experience into...
The performance of a language model has been shown to be effectively mod...
We enhance auto-regressive language models by conditioning on document c...
Sparse neural networks are becoming increasingly important as the field ...
Whilst there are perhaps only a few scientific methods, there seem to be...
We address the problem of building theoretical models that help elucidat...
We attempt to automate various artistic processes by inventing a set of
...
Highly overparametrized neural networks can display curiously strong
gen...
Neural networks have historically been built layerwise from the set of
f...
Current methods for training recurrent neural networks are based on
back...
The work "Loss Landscape Sightseeing with Multi-Point Optimization"
(Sko...
Determining what experience to generate to best facilitate learning (i.e...
We study the problem of learning associative memory – a system which is ...
In this report we review memory-based meta-learning as a tool for buildi...
The transfer of knowledge from one policy to another is an important too...
Gradient-based meta-learning techniques are both widely applicable and
p...
We introduce a class of causal video understanding models that aims to
i...
The scope of the Baldwin effect was recently called into question by two...
We introduce Mix&Match (M&M) - a training framework designed to facilita...
We present a method for using previously-trained 'teacher' agents to
kic...
Neural networks dominate the modern machine learning landscape, but thei...
User participation in online communities is driven by the intertwinement...
When training neural networks, the use of Synthetic Gradients (SG) allow...
Training directed neural networks typically requires forward-propagating...
We present a novel deep recurrent neural network architecture that learn...
We present recursive recurrent neural networks with attention modeling
(...
We propose a simple and straightforward way of creating powerful image
r...
Generative Adversarial Nets [8] were recently introduced as a novel way ...