What I cannot create, I do not understand. - Richard Feynman
Reproduction of DINOv1.
Reproduction of SegFormer.
Reproduction of Detection Transformer.
Reproduction of ViT, DeiT, Swin and MobileViT.
Reproduction and attempt at pretraining GPT-2.
Reproduction of denoising diffusion probabilistic models.
Reproduction of SimGAN in TensorFlow 2.
Minimal YOLOv11 inference wrapper using OpenCV DNN with optimized NMS.
CNN for detecting facial landmarks with head pose estimation.
Tiny autograd engine similar to PyTorch, with tensor and NN APIs.
Minimal implementation of Byte Pair Encoding used in LLM tokenization.
A friend's business site that I built and manage.
Minimal and lightweight Hugo theme with markdown-ish aesthetics.