Training neural nets, building web apps, and refusing to pick a lane.
Moved here from medium.com/@geronimo7.
How FLUX.2 Klein Generates Images
Text encoder, VAE and denoiser code built from scratch one component at a time
Chest X-Ray Classification with DINOv3
Training a Neural Network to Spot Disease
Build the Most Simple CSV RAG From Scratch
Step by step. 100 lines of Python without the bloat
Client-Side NSFW Image Detection with DINOv3
Train a classifier with Meta’s new model and run it in the browser
Client-Side Image Inpainting with ONNX and Next.js
LaMa in the Browser. It's not the llama you know
Training a Latent Diffusion Model From Scratch
A Hands-On Journey. We start from 0 with a desktop GPU only
Multi GPU Training with PyTorch
Getting Started with Distributed Data Parallel (DDP)
VLM Inference on multiple GPUs
Parallel Image Captioning with SmolVLM2, Moondream2, Qwen2.5-VL and DDP
How NVIDIA's Sana Generates Images
Autoencoder, diffusion transformer, and the leap to 4K
Smart Background Removal in the Browser
A depth-based image editor with transformers.js and Depth Anything
Image Segmentation in the Browser with Segment Anything Model 2
Run Meta’s SAM2 with onnxruntime-web
Photo Editing with AI
Inpainting with Stable Diffusion and FLUX
Train a FLUX.1 LoRA for $1
Step-by-Step Guide to Training on X-Rays (or any kind of Image)
Evaluating LLMs with Semantic Similarity
Testing SemScore to evaluate LLM output
LLM Inference on multiple GPUs with 🤗 Accelerate
Minimal working examples and performance benchmark
Finetuning Llama 2 and Mistral
A beginner’s guide to finetuning LLMs with QLoRA