make a shellgpt4all

[luisfilipesilvasantos]

can you take the projet in https://pypi.org/project/shell-gpt/ and make one that uses gpt4all-falcon that will not need to use the open ai api key?

[cebtenzzre]

This is an interesting suggestion, but I think it's really up to the community to make something like this, as it has very different goals than GPT4All. Perhaps it could be built on our Python bindings.

llama-cpp-python provides a mature OpenAI-compatible server that you could use as a drop-in replacement for OpenAI's GPT with shell-gpt: ggml-org/llama.cpp#795

Then you can use gpt4all-falcon as the model if you'd like.