Installing Ollama
MacOS:
1
brew install ollama
Linux:
1
curl -fsSL https://ollama.com/install.sh | sh
Install official models
1
ollama pull <modelname>
Then use run
use the model
Installing custom huggingface models in ollama
First find a .gguf model file and download it.
Then create a modelfile (the filename does not matter)
1
2
3
4
5
6
7
8
9
10
11
12
13
FROM /path/to/model.gguf
TEMPLATE """<|im_start|>system
<|im_end|>
<|im_start|>user
<|im_end|>
<|im_start|>assistant
"""
SYSTEM """"""
PARAMETER stop "<|im_start|>"
PARAMETER stop "<|im_end|>"
Then run the following command to install the model:
1
ollama create PreferedModelName -f /path/to/modelfile