# Educational LLMs (GGUF format)

Small openly-released language models in GGUF format,
suitable for running on consumer hardware. Useful for
studying inference techniques and quantization methods.


## Contents

| Model | Size | RAM needed | Notes |
|-------|------|------------|-------|
| TinyLlama-1.1B Q4 | 700MB | ~1GB | Apache-2.0; surprisingly coherent for size |
| Phi-2 Q4 | 1.5GB | ~2GB | Microsoft's small research model |

## Usage with llama.cpp

    cd llama.cpp && make
    ./main -m /path/to/model.gguf -p "Once upon a time"

## Notes

These are quantized variants of openly-released models, kept
locally for offline experimentation. Refer to each model's
original release notes for licensing and intended use.