A book that builds GPT-2, Llama 3, DeepSeek from scratch in PyTorch
Loads real Meta and OpenAI weights, not just training from scratch.
Companion code for Machine Learning From Scratch — 10 core ML algorithms built from scratch with NumPy, compared with Scikit-learn and PyTorch.
Yet another ML-from-scratch teaching repo in a crowded educational space.
Students learning ML fundamentals, self-taught developers
grokking-deep-learning · ML-From-Scratch · fastai
Loads real Meta and OpenAI weights, not just training from scratch.
Hopper and Google Flights already dominate this space with better data.
Elegant single-problem solver for second-hand book hunters, but narrow audience.
Claims new DNN+LLM architecture for deterministic OCR and scraping tasks.
Multi-exchange prediction market terminal with arbitrage scanner beats tab-switching chaos.
Five-agent LangGraph pipeline beats single-prompt tools like Simplify and Teal.