vim-llama

vim-llama is a vimPlug plugin for self hosted codellama completion

Requirements

You will need to have docker as well as vim installed.

Install

ollama local server

Pull the ollama docker image

docker pull ollama/ollama

And the run it on the port 11434 in a container named ollama:

docker run -d -v ollama:/root/.ollama -p 11434:11434 --name ollama ollama/ollama

Note that you will have to pull the model you want to use using the VLMAPull <model> command before using it or an error will be thrown.

vimplug plugin

Open your .vimrc file and add the following code between the call plug#begin('~/.vim/plugged') and call plug#end() statements:

Plug 'Dr4x14913/vim-llama'

Then open vim and run:

:PlugInstall

Configuration

You can tune the following variables:

Name	Default value	What this is
g:vim_llama_context_size	20	Size of the context window: number of lines above your cursor that will be taken into account when start function is called. If you are in v mode, only selected lines will be taken into account.
g:vim_llama_model	codellama	Model that you want to use (ex: llama2, codellama:python ...)
g:vim_llama_ip	localhost	IP address of the Llama server
g:vim_llama_port	11434	Port number for the Llama server

Use it

You can type :VLMAStart command and the model will continue your code based on the g:vim_llama_context_size lines before the one you are on at this time.
You can alose type this command in visual mode, in this case, only the selected lines will be passed to the model.

The default behaviour of the model will be to continue your code, if you want to do different stuff, you can run:

:VLMAStart <Begining instructions>, <End instructions>

This will prompt:

<Begining instructions>
```
# Lines that you selected in visual mode
# or lines that are in the context window
```
<End insructions>

Available commands

Command name	Option	Behaviour
VLMAStart	[<range>], [<prompt start>], [<prompt end>]	Starts the completion as explained above.
VLMAStop	-	Stops a running completion process.
VLMAPull	model	Pulls an ollama model.
VLMALogs	-	Displays the logs of the last completion.