qeep

I’ve spent the last two years building qeep, a deep learning framework written entirely in Go. Why I built this: While Go has some great math libraries, I wanted a framework that felt more like PyTorch (declarative API) but stayed true to Go’s simplicity. I also wanted to deeply understand how AutoGrad and CUDA kernels interface with Go’s runtime and memory management. Key Features: - Multi-Dimensional Tensors with a wide range of linear algebra and statistical operations. - Automatic differentiation (AutoGrad) for tensors. - GPU acceleration via CUDA for high-performance large tensor computations. - A variety of neural network components, such as fully connected (FC) layer. - A declarative API for defining neural networks using stream package. Repo: https://github.com/sahandsafizadeh/qeep