Topic

Large language models

Open-weight and commercial large language models, with the APIs, fine-tuning methods, prompting techniques and guides for putting them to work.

Large language models (LLMs) are neural networks trained on large amounts of text to predict the next token, which lets them answer questions, write and edit text and follow instructions. To start with large language models, run one on your own machine with the guide to Ollama, or call a hosted model through the guide to structured outputs with JSON schema in the OpenAI and Claude APIs.

The tool pages cover local runners such as Ollama and LM Studio, Gemini Notebook for questions over your own documents, and Langfuse for tracing LLM apps. Open-source project pages focus on cutting the tokens sent to a model, plus a free course that teaches AI systems starting from linear algebra. A third guide covers serving an open model with vLLM behind an OpenAI-compatible API.

The videos range from a 3Blue1Brown explainer on cross-entropy to talks on inference speed, retrieval versus fine-tuning and production monitoring. Each video card shows length and level, so you can pick a short introduction or a longer talk.