privateGPT is a question-and-answer platform for privatized deployment documents based on GPT4All-J. It does not need to be connected to the Internet and can 100% guarantee that users’ privacy will not be leaked.

It provides an API where users can interactively question answer and generate text using their own documents. In addition, the platform supports custom training data and model parameters to meet individual needs.

The privateGPT technology stack is LangChain and GPT4All

• LLM uses ggml-model-q4_0.bin by default
• Embedding uses ggml-model-q4_0.bin by default