Alex Redden

flux-fp8-api. Flux diffusion model implementation using quantized fp8 matmul & remaining layers use faster half precision accumulate, which is ~2x faster on consumer devices.

electron-G14Control. Electron desktop application replacement for G14ControlR3.

torch-cublas-hgemm. PyTorch half precision gemm lib w/ fused optional bias + optional relu/gelu

torch-bnb-fp4. Faster Pytorch bitsandbytes 4bit fp4 nn.Linear ops

G14ControlR3. Windows tray application which allows for control over processor boost mode, fan speed, CPU TDP, and Windows power plans for the ASUS Zephyrus G14 with Ryzen 4000 series CPUs.

RestartGPU. Simple command line app for restarting a single nvidia gpu

denoisemoji. command-line tool that allows you to create a pseudo diffusion denoised emoji gif from a PNG/WEBP/JPG image.

f8_matmul_fast. Cuda

torch_xxhash. C++

SAC-dataset-knn-dl-query. Repo with simple scripts for downloading the SAC dataset via multiprocessing, creating image embeddings multi-gpu, creating a KNN index, and then querying the knn dataset & viewing returned image grids.

object-detection-display. Fast multithreaded object detection with yolov5 and a real-time opencv image display.

interrogator. Python

functis. Universal image reader designed for seamless reading of images into multiple formats, including PIL Images, Torch Tensors, and NumPy arrays.

faster-cogagent-chat. Python

stable-deforum-diffusion. Jupyter Notebook

energy-monitor-server. Python

pydotted. A python dictionary that can get attribute (x.y) access, including nested dicts.