Model training

Adjusting a model’s parameters using data so it learns patterns for a task.

Training changes the model itself. Inference uses a model that has already been trained. Compact projects can make that distinction easier to explore.

Related primary source

Projects connected to this idea

nanoGPT

Minimal code for training and fine-tuning GPT-style language models.

Project notesLearn MLCode library

llm.c

Explore language-model training through a compact C and CUDA implementation.

Project notesLearn MLC / CUDA code

nanochat

A compact toolkit spanning language-model training, evaluation and inference.

Project notesLearn MLTraining toolkit

autoresearch

Let coding agents run and compare bounded model-training experiments.

Project notesLearn MLResearch workflow