Listening to the JS Party episode about Jerod Santo's tool box motivated me to finally get Ollama working properly.

I had already played around with Ollama when I gave my friend Slava's datachat a whirl. In that case, per the README, I spun up a Docker image and ran the following to pull the 4-5 GB model in:

docker exec -it ollama /bin/bash
ollama pull llama3

Then Slava's app, which is pretty neat by the way, worked just fine.

In the podcast, Jerod was talking specifically about Enchanted, so that was my starting point. The README was a little light on how to install Ollama but the Ollama website itself was very clear. On Mac, it's just a zip file you unpack and drag into your “Applications” folder.

I was prompted to install the Ollama CLI, which I did, and it told me how to download and run a model:

ollama run llama3.2

To check if Ollama's working, you could try this, from the Ollama README:

curl http://localhost:11434/api/generate -d '{
  "model": "llama3.2",
  "prompt":"Why is the sky blue?"
}'

You should get a bunch of JSON back.

Ollama is just a backend and they have a long list of frontends in their README, including Enchanted.

Enchanted ended up not working well for me. As I mentioned in the episode's comments, I can't simply type messages over and over again in the prompt. Instead, I have to click in the prompt each time I want to type again. There are a couple issues about this.

Looking for an alternative UI, I went back to the list and picked the one at the top, Open Web UI. It worked fine but I do like me a fat client.

Over the last few months, I've been playing with a new editor called Zed and I know that two of its core features are that it's multiplayer and AI ready. Here's how I configured Zed for Ollama:

• At the bottom right, click the “Assistant” button. It looks like a star.
• At the top right, click the hamburger button and then “Configure”.
• On the “configure your assistant” screen, scroll down until you see Ollama. It should say “connected” and above this, click “open new context”.
• Now you should be at a prompt and can start sending messages to Ollama.

I've played with ChatGPT, of course, but it's fun that Ollama is completely free, open source, and runs locally on my laptop.