First Principles of Evaluating Deep Learning Inference

Numerical Errors are Evil

Attention with Learned Spatial Feature Sampling

Convolution with Learned Spatial Feature Sampling

Using Motorola CPU32 as an Example

How ONNX Expands the Support for Deep Learning Inference Over the Years

Understanding PyTorch Leaf Tensor

Personal Unwarranted ONNX Contribution Guideline

ONNX Runtime Python Inference Example

Generative Pre-Trained Transformer Models From OpenAI

Principles for Faster Transformer Inference

Accelerated Low Bit-Width Neural Network Training