Following the tutorial and this. For optimization PEFT

• Install miniconda
• Start Anaconda in your terminal. For example, open Alacritty and run C:\Users\neupo\miniconda3\Scripts\activate.bat. Also see VS Code example.

conda create -n starchat python=3.10
conda activate starchat

git clone https://github.com/bigcode-project/starcoder.git
cd starcoder/chat
pip install -r requirements.txt

To exit the environment type

conda deactivate

Create dataset

Comparison with GPT and supposedly heuristics on how to prepare a dataset from github repositories (paper). Big Query example and another