- Install dependencies with:
pip install -r requirements.txt- Run the script
download_model.shby:
sh download_model.sh 117M
This should download the gpt-2 model. 117M is the smallest model, 345M is a larger variant.
- Create
auth.json, and place it inside theconfigfolder. Its content should be:
{
"token": "<your_token>",
"client_id": "<client_id>"
}- Run the script with:
python gpt-chatbot-client.py- Enable finetuning.