llama-wiiu.cpp

This is just a quick little vibe-coded project I made. The model I tested with was 'qwen2.5-0.5b-instruct-q4_0.gguf' Currently the prompt has to be hardcoded before building.

### Instruction
You are a helpful assistant. Reply conversationally and directly.
### Input:
Hi
### Response:

To edit the "Hi" and other settings here in wiiu/main.cpp

//CONFIG ######################
const int n_predict = 32;
const std::string textInput = "Hi!";
const bool use_buggy_keyboard = false; 
//#############################

Running on hardware

• Copy a model (.gguf) to sd:/model/ (Don't put multiple in there).
• Run llama-wiiu.wuhb.
• A log is also written to sd:/llama-wiiu.log.

Build

Requirements:

• devkitPro with WUT

Build the RPX/WUHB:

make

Outputs:

• build-wiiu/llama-wiiu.rpx
• build-wiiu/llama-wiiu.wuhb