Abstract Machine learning models are increasingly being scaled in both sequence length
and model dimension to reach longer contexts and better performance. However, existing …

The end-to-end performance of deep learning model inference is often limited by excess
data movement on GPUs. To reduce data movement, existing deep learning frameworks …

Optimization on tensor computations can significantly reduce inference and training time;
therefore, tensor computations represent the fundamental computational bottleneck in …

The advanced data modelling capabilities of neural networks allowed deep learning to
become a cornerstone of many applications of artificial intelligence (AI). AI can be brought …