Github Copilot but for Overleaf.

A Chrome extension and finetuning script to run a local LLM latex completion.

Getting Started

1. Clone the Repository

   git clone https://github.com/adityamakkar000/OverleafCoPilot.git

2. Setup Env

   Setup the python environment using

   pip install -r req.txt

3. Finetune

   Place your latex .txt in ai/dataset/ and create a .yaml file in ai/configs/ following the structure in ai/configs/main.yaml Run

   python3 finetune.py --config-name $config-name arg1=$arg1 arg2=$arg2 ...

4. Use

   Run a local server using

   python3 main.py path=$(path to finetuned model) tokens=$tokens

   Add the ip-address in extension/sentence.js and load the extension folder in chrome://extensions/ Add ip-address at chrome://flags/ to allow for mixed-content on http and https Reload chrome and press alt to run inference