Android front end for inference on a remote server, with optional logging and speech

[compilebunny]

AndroidRemoteGPT is an android front end for inference on a remote server using open source generative AI models.

Most Android devices can't run inference reasonably because of processing and memory limitations. The next best thing is to run the models on a remote server but access them through your handheld device. AndroidRemoteGPT allows you to send queries and get responses on your phone, given that you have a server running a model somewhere.

This pre-release is quite basic. Plans include:

• Pretty up the interface
• Add an icon so that AndroidRemoteGPT can be launched from Android directly without first loading Termux
• Add an on-device inference option for people who have 8gb of RAM on their android devices
• Allow ssh passwords?

If you try it, please tell me what you think and what I should add next.