By Greninja9257
A native macOS lab for teaching tiny language models to think β build the architecture, train the weights, and watch a small LLM emerge from scratch, locally on Apple Silicon with custom data, tokenizers, checkpoints, and MLX acceleration.