Description
I know that you have closed a previously opened issue about that topic on December 14th due to lack of answer from the original poster but meanwhile more potent smartphones are making their headway, so the issue is very relevant again ;-)
Here is h a copy of the message I left on your closed thread.
Fantastic project, kudos!
I'm also interested in running a choice of models from an Android smartphone.
I'm a developer and my use case involves an app driving the phone by voice (prompts will be acquired by a speech to text fronted) triggering agents from a bound LLM (supporting function calls).
I just bought a state of the art phone just to make this possible, it's a OnePlus 12 featuring a Snapdragon 8 gen 3 SOC with 24Gb RAM (available only from China!) and 1tb storage. It should be enough to execute quantized 13B models .
My app will be created with Flutter and should interoperate with the picked llamafile model via Dart FFI.
Activity