https://github.com/ivanfioravanti/autogram

Grammar checker with a keyboard shortcut for Ollama and Apple MLX with Automator on macOS.
https://github.com/ivanfioravanti/autogram

Last synced: about 2 months ago
JSON representation

Grammar checker with a keyboard shortcut for Ollama and Apple MLX with Automator on macOS.

• Host: GitHub
• URL: https://github.com/ivanfioravanti/autogram
• Owner: ivanfioravanti
• License: mit
• Created: 2024-02-04T18:21:17.000Z (11 months ago)
• Default Branch: main
• Last Pushed: 2024-02-05T15:41:39.000Z (11 months ago)
• Last Synced: 2024-11-07T03:04:53.942Z (2 months ago)
• Language: Python
• Homepage:
• Size: 946 KB
• Stars: 76
• Watchers: 3
• Forks: 5
• Open Issues: 1
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Autogram



Grammar checker for Apple Silicon devices powered by Automator with Ollama or Apple MLX as backend.

Open, free, easy and fast! 

Play, copy, fork, experiment, have fun!

## Ollama Backend

### Installation

Install [ollama](https://ollama.ai/)

Run in the background to download a Mistral Instruct 0.2 based model:

```bash

ollama run ifioravanti/mistral-grammar-checker

```

Note: If you have base model installed in Ollama, only a very small additional layer will be downloaded.

You can easily create new custom models using ModelFiles [following instructions on Ollama](https://github.com/ollama/ollama/blob/main/docs/modelfile.md)

Install the workflow. This essentially copies the workflow to `~/Library/Services/autogram-ollama.workflow`.





Go to System Preferences -> Keyboard -> Shortcuts -> Services -> General -> autogram, and set the shortcut, in my case I set it to `⌃⌥⌘G`.





### Usage

In any OSX application, select some text, and press the shortcut you set. The selected text will be replaced with the generated text from the model.

## Apple MLX Backend

Install `autogram-mlx.workflow` the same way as above, no need to edit and add token, however you may edit `MODEL="mistral"`, in case you want to use a different model



This worfklow is based on mlx-community/Mistral-7B-Instruct-v0.2-4bit-mlx model with prompt defined directly in the workflow.

You can edit it directly from Automator.

### Credits

Original idea [LLM-Automator](https://github.com/radames/LLM-automator) [Radamés Ajna](https://github.com/radames)