Ollama on Windows: setup and Gemma 3 tutorial

This tutorial demonstrates installing Ollama on Windows and chatting with Gemma 3 1B through its graphical interface and terminal. The speaker explains that local use can work offline after downloading the installer and model files, which requires internet access.

The walkthrough recommends 8 GB of RAM and shows how to inspect memory in Task Manager. The speaker suggests models with 1 to 3 billion parameters for an 8 GB machine and models around 7 to 8 billion parameters for a 16 GB machine. These are the speaker's sizing guidelines, with a reminder to leave memory for Windows.

The demonstration downloads Gemma 3 1B, listed at 815 MB, using Ollama's pull command. The speaker checks the download with list, reopens the graphical interface and selects the model. He then adjusts context length and requests an essay about AI use for homework. The narration gives an unclear unit for the context setting, so it does not establish a precise recommended value. The speaker also notes that response speed depends on the computer.

The final section shows the model storage directory and explains how to change it. Terminal examples cover starting a chat with run, ending the conversation, listing downloaded models and removing a model with rm.