32dots HEIDELBERG AI
Curriculum › Ollama course › ollama-01
Session 1 easy

Pull, manage & compare models from the CLI

USE 0 - 20 min

Learn the handful of commands that manage your local model library

Ollama is CLI-first, and almost everything you need is a few short commands. ollama pull downloads a model without starting a chat; ollama list shows what you have; ollama ps shows what is currently loaded in RAM; ollama stop frees that RAM; and ollama rm deletes a model from disk. Knowing these five turns Ollama from a chat toy into a managed local model library you can keep tidy on a laptop with limited disk space.

  1. 1 Pull a model without chatting: run ollama pull qwen2.5. This downloads it so it is ready instantly later.
  2. 2 List what you have: run ollama list. You will see each model's name, size on disk, and when it was modified.
  3. 3 See what is loaded in memory: run a model (ollama run llama3 in another terminal), then run ollama ps. A loaded model is using RAM even between questions.
  4. 4 Free that RAM: run ollama stop llama3. Confirm with ollama ps that it is no longer loaded.
  5. 5 Delete a model you no longer need: run ollama rm to reclaim the disk space. Check with ollama list again.
✓

You can list, load, stop, and remove models from the command line, and you understand which command frees RAM versus disk.

BUILD 20 - 30 min

Build a two-model verdict for a real research question

Run a genuine question from your current work through two models and produce a short comparison note — the kind you might share with a lab-mate before choosing a tool.

Your task

Pick a real question from your research (a methods choice, a literature gap, an analysis approach). Ask two different models. Write two sentences on which you would trust more and why.

  1. 1 Choose a question you actually need an answer to — not a toy example. For instance: I have 12 RNA-seq samples across 3 conditions. What statistical approach should I use to find differentially expressed genes, and what are the main pitfalls?
  2. 2 Run it through ollama run , then through ollama run (a different family).
  3. 3 For each response, mark one strength and one weakness.
  4. 4 Write your two-sentence verdict: model A vs model B, and which you would use for this task.
Deliverable

The prompt you used, both responses (paste or screenshot), and your two-sentence verdict.