Skip to content

Repository files navigation

Mobile LLM Stack

An experimental end-to-end on-device LLM platform for constrained Android devices.

Research question

Can a constrained Android device support an end-to-end local LLM lifecycle, from parameter-efficient fine-tuning through deployment and resource-aware inference?

v0.1 demonstrated path

MobileFineTuner → LoRA fine-tuning → checkpoint creation → native model merge/export → native C++ SafeTensors → GGUF conversion → llama.cpp → Android / Termux inference

Components

MobileFineTuner

Training plane for native C++ on-device model adaptation.

model-pipeline

Model artifact validation, checkpoint inspection, merging/export, and native GGUF conversion.

MOBILE-INFER

Inference/runtime plane for persistent and resource-aware mobile LLM execution.

Current demonstration

GPT-2 Small:

  • 12 transformer layers
  • 768 hidden dimension
  • 12 attention heads
  • 124M parameters
  • LoRA rank 8
  • WikiText-2 fine-tuning
  • Native F32 GGUF deployment
  • llama.cpp inference on Android / Termux

Observed generation performance on the test Android environment was approximately 10–12 tokens/second for F32 GPT-2 Small.

Status

v0.1 is an experimental research prototype.

The current release demonstrates end-to-end execution but does not yet claim numerical equivalence against a reference GGUF conversion.

Related work

This project builds on and integrates existing work including:

  • MobileFineTuner
  • llama.cpp
  • GPT-2
  • LoRA
  • SafeTensors
  • GGUF
  • WikiText-2
  • prior research on mobile LLM inference and memory management

This repository does not claim ownership of those underlying technologies.

License

Apache-2.0

About

End-to-end on-device LLM training, conversion, deployment and inference for constrained Android devices

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages