optimization
- Streaming ConvolutionsAugust 24, 2023
Working out the math for streaming convolutions. - Fast Attention ImplementationsJune 29, 2023
A reference collection of fast attention implementations. - RWKV Language Model MathJune 16, 2023
In-depth explanation of the math behind the RWKV model, with PyTorch implementations, plus a discussion of numerical stability. - Optimized Log-Sum-Exp PyTorch FunctionMay 20, 2020
A walkthrough of how to optimize the log-sum-exp function in PyTorch. - Torch CUDA Extension TricksMay 6, 2020
Some tricks I found useful for writing CUDA extensions for PyTorch.