echodict/llama.cpp

1## gguf2 3This is a Python package for writing binary files in the [GGUF](https://github.com/ggml-org/ggml/pull/302)4(GGML Universal File) format.5 6See [convert_hf_to_gguf.py](https://github.com/ggml-org/llama.cpp/blob/master/convert_hf_to_gguf.py)7as an example for its usage.8 9## Installation10```sh11pip install gguf12```13 14Optionally, you can install gguf with the extra 'gui' to enable the visual GGUF editor.15```sh16pip install gguf[gui]17```18 19## API Examples/Simple Tools20 21[examples/writer.py](https://github.com/ggml-org/llama.cpp/blob/master/gguf-py/examples/writer.py) — Generates `example.gguf` in the current directory to demonstrate generating a GGUF file. Note that this file cannot be used as a model.22 23[examples/reader.py](https://github.com/ggml-org/llama.cpp/blob/master/gguf-py/examples/reader.py) — Extracts and displays key-value pairs and tensor details from a GGUF file in a readable format.24 25[gguf/scripts/gguf_dump.py](https://github.com/ggml-org/llama.cpp/blob/master/gguf-py/gguf/scripts/gguf_dump.py) — Dumps a GGUF file's metadata to the console.26 27[gguf/scripts/gguf_set_metadata.py](https://github.com/ggml-org/llama.cpp/blob/master/gguf-py/gguf/scripts/gguf_set_metadata.py) — Allows changing simple metadata values in a GGUF file by key.28 29[gguf/scripts/gguf_convert_endian.py](https://github.com/ggml-org/llama.cpp/blob/master/gguf-py/gguf/scripts/gguf_convert_endian.py) — Allows converting the endianness of GGUF files.30 31[gguf/scripts/gguf_new_metadata.py](https://github.com/ggml-org/llama.cpp/blob/master/gguf-py/gguf/scripts/gguf_new_metadata.py) — Copies a GGUF file with added/modified/removed metadata values.32 33[gguf/scripts/gguf_editor_gui.py](https://github.com/ggml-org/llama.cpp/blob/master/gguf-py/gguf/scripts/gguf_editor_gui.py) — Allows for viewing, editing, adding, or removing metadata values within a GGUF file as well as viewing its tensors with a Qt interface.34 35## Development36Maintainers who participate in development of this package are advised to install it in editable mode:37 38```sh39cd /path/to/llama.cpp/gguf-py40 41pip install --editable .42```43 44**Note**: This may require to upgrade your Pip installation, with a message saying that editable installation currently requires `setup.py`.45In this case, upgrade Pip to the latest:46 47```sh48pip install --upgrade pip49```50 51## Automatic publishing with CI52 53There's a GitHub workflow to make a release automatically upon creation of tags in a specified format.54 551. Bump the version in `pyproject.toml`.562. Create a tag named `gguf-vx.x.x` where `x.x.x` is the semantic version number.57 58```sh59git tag -a gguf-v1.0.0 -m "Version 1.0 release"60```61 623. Push the tags.63 64```sh65git push origin --tags66```67 68## Manual publishing69If you want to publish the package manually for any reason, you need to have `twine` and `build` installed:70 71```sh72pip install build twine73```74 75Then, follow these steps to release a new version:76 771. Bump the version in `pyproject.toml`.782. Build the package:79 80```sh81python -m build82```83 843. Upload the generated distribution archives:85 86```sh87python -m twine upload dist/*88```89 90## Run Unit Tests91 92From root of this repository you can run this command to run all the unit tests93 94```bash95python -m unittest discover ./gguf-py -v96```97 98## TODO99- [ ] Include conversion scripts as command line entry points in this package.100