# (NOT YET TESTED COMPLETELY) Run local AI models on your Samsung Galaxy Note 20 Ultra 5G with Snapdragon 865+ (Adreno 650 GPU) via Termux.

locally.md

Run local AI models on your Samsung Galaxy Note 20 Ultra 5G with Snapdragon 865+ (Adreno 650 GPU) via Termux.

---

1. Install Termux and Termux:API

Install Termux from F-Droid or GitHub.(Davide Fornelli)

Then, install Termux:API:(Davide Fornelli)

pkg install termux-api

---

2. Set Up Termux Storage Access

Grant Termux access to your device's storage:(GitHub)

termux-setup-storage

---

3. Install Dependencies

Update packages and install required dependencies:(GitHub)

pkg update && pkg upgrade
pkg install git clang cmake make python python-dev libomp wget
pkg install vulkan-tools
pkg install termux-api

---

4. Clone and Build llama.cpp

Clone the llama.cpp repository and build it:(Mendhak Code)

git clone https://github.com/ggerganov/llama.cpp
cd llama.cpp
make clean
make -j$(nproc)

---

5. Download a Model

Download a quantized GGUF model (e.g., 7B) from a trusted source and place it in the llama.cpp directory.

---

6. Run the Model

Execute the model using the built binary:(Mendhak Code)

./main -m model.gguf -p "Hello, AI!"

---

7. Enable GPU Acceleration (Optional)

To utilize GPU acceleration:(Wolfchip Electronics)

1. Install Vulkan Drivers:

   Ensure Vulkan drivers are installed on your device.

2. Build with Vulkan Support:

   Modify the build process to include Vulkan support.

   make clean
   make USE_VULKAN=1 -j$(nproc)

3. Run with Vulkan:

   Execute the model with Vulkan support:

   ./main -m model.gguf -p "Hello, GPU!"

---

Resources

• Guide to Run Local AI on Android with llama.cpp and Termux
• llama.cpp Android Tutorial by JackZeng0208
• Android-Termux-LLM-Tutorial by Jie-Qiao

---

This setup allows you to run local AI models on your Android device using Termux, leveraging the Adreno 650 GPU for accelerated inference.