White Paper

A Path Towards Autonomous Machine Intelligence

Design Principles for Lifelong Learning AI Accelerators

A Review of Sparse Expert Models in Deep Learning

MegaBlocks: Efficient Sparse Training with Mixture-of-Experts

The Reversible Residual Network: Backpropagation Without Storing Activations

UDC: Unified DNAS for Compressible TinyML Models for Neural Processing Units

Predictive Coding Towards a Future of Deep Learning Beyond Backpropagation

Reversible Architectures for Arbitrarily Deep Residual Neural Networks

Decoupled Neural Interfaces using Synthetic Gradients

Attention Is All You Need