1 dépôt
Optimized GPU kernels that fuse projections and rotations specifically for transformer layers.
Distinct from Fused GPU Kernel Composition: Specializes in fusion within transformer projections, whereas Fused GPU Kernel Composition is a general technique for sequential operations.
Explore 1 awesome GitHub repository matching artificial intelligence & ml · Transformer Projection Kernels. Refine with filters or upvote what's useful.
ComfyUI-nunchaku is a 4-bit diffusion inference engine and a set of nodes for running low-precision quantized diffusion models within ComfyUI visual workflows. It provides a backend that reduces memory overhead and increases generation speed for transformer models. The project includes specialized tools for identity-preserving generation and an image-to-image guidance toolkit that uses depth maps and reference images. It also features a multimodal visual question answering implementation and a utility for merging multiple quantized model files into single unified files. The engine covers a b
Accelerates inference by combining projections and rotations into single optimized kernels within transformer layers.