Skip to content

CLI to run huggingface models on MLX. Check readme for model support

Notifications You must be signed in to change notification settings

vishalvrv9/emex

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

9 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

emex

Emex helps you run a chat with an LLM locally, via a terminal, by leveraging Apple's optimized MLX array framework.

Demo Video


Usage


Future plans to integrate with a package manager (brew/pip)

First clone this repo & cd into project root

  • create a python environment (preferred)
python3 -m venv create .env
source .env/bin/activate

Install the dependencies

pip install .

Simply run using

emex run username/modelname

Here, username/modelname refers to the huggingface model you choose to run locally. To browse the list of compatible models, you can browse all the models on HugginFace under the mlx-community here


Currently Supported Models:

  • microsoft/phi-2
  • gemma-2b

All models currently used are using models from the mlx-community. Other models from mlx-community within hugginface should work out of the box but are yet to be tested.

The cli uses mlx_lm and mlx to generate text


Features

  • Generate text using local LLMs via MLX
  • Verbose mode to print metrics like Tokens/Sec, (TTFS) Time to first token & Total Time taken for generation

About

CLI to run huggingface models on MLX. Check readme for model support

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages