# Woteale Qwen Prompt Pack v1

This is a small configuration package for the official
`Qwen/Qwen2.5-0.5B-Instruct` base at revision
`7ae557604adf67be50417f59c2c2f167def9a775`.

## What changes

Only the system prompt and runtime configuration. No Qwen weights are included,
trained, merged, quantized, or redistributed. This package must not be
described as a Woteale-trained model or as a performance improvement.

## Load

```bash
pip install "transformers>=4.37.0" accelerate torch
python load_woteale.py
```

The command downloads the official Apache-2.0 base model from Hugging Face on
first use. Review `NOTICE.md` and the base-model licence before distribution.

## Files

- `woteale-system-prompt.txt`: Woteale's Japanese system prompt.
- `woteale-qwen-config.json`: fixed base model/revision and runtime metadata.
- `load_woteale.py`: minimal `transformers` loader.
- `NOTICE.md`: base-model attribution and modification boundary.