This repo contains pretrained models and tools that are to be used for modelling human-aligned dialog.
| Model ID | Parameter Count | Training Method |
|---|---|---|
| convogpt-125m-uft | 125 Million | Unsupervised Finetuning |
| convogpt-1.3b-uft | 1.3 Billion | Unsupervised Finetuning |
| convogpt-2.7b-uft | 2.7 Billion | Unsupervised Finetuning |
| convogpt-6b-uft | 6 Billion | Unsupervised Finetuning |
git clone [email protected]:harubaru/convogpt.git
cd convogpt
python -m pip install -r requirements