Run Ollama locally

I use Ollama to run my custom patterns locally (so I do not have to pay for an OpenAI API key). It is slower but fun to use. If you want to utilize it with the Fabric tool, you must run this inside your WSL instance.

Run curl -fsSL https://ollama.com/install.sh | sh in your terminal to download and run the Ollama server.

Use ollama help to list available commands.

Run ollama run llama3.1:8b to pull the latest Llama3 model from Ollama.

If you are currently running Fabric, you can use fabric --listmodels to see all available models for the application.

Example command: echo "What is the meaning of life?" | fabric --pattern ai --model llama3.1:8b --stream

Boom, you are in business.

If you would like to take this a step further, you can create an alias in bash so you do not have to type out the entire command all the time.

For example, I use xclip to copy and paste content into Fabric. But instead of typing out xclip commands, I would just like to use pbpaste. Try this:

1. Open your .bashrc file: nano ~/.bashrc

2. Download xclip using apt: sudo apt install xclip

3. Copy and paste alias pbpaste='xclip -sel clip -o' at the bottom of your file.

4. Run pbpaste in your terminal to see results from your clipboard.

Taking it one step further:

alias ai='fabric --pattern ai --model llama3.1:8b --stream'

echo 'What is the meaning of life?' | ai

References:

libraryollama

Ollama available models