tldr: LoRA fine-tuning of an open-source alpaca model.

See https://arxiv.org/abs/2305.15717 for why calling this a private ChatGPT is probably misleading.