Metadata-Version: 2.4
Name: uni-quant-cuda
Version: 0.2.0
Summary: Uni-Quant: CUDA-accelerated quantization/dequantization for Keras and XGBoost models
Author-email: Jakub Grula <ramsters110@gmail.com>
License: MIT
Classifier: Programming Language :: Python :: 3
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: POSIX :: Linux
Requires-Python: >=3.11
Description-Content-Type: text/markdown
License-File: LICENSE
Dynamic: license-file

Uni-Quant
========

Small library to quantize/dequantize Keras and XGBoost models using PyTorch CUDA kernels.

Notes
- This package compiles CUDA kernels at runtime using `torch.utils.cpp_extension.load_inline`.
- Installing and using the CUDA compilation requires a compatible PyTorch build and CUDA toolkit on the target machine.

Dependencies are listed in `requirements.txt` and synchronized with `pyproject.toml`.

Quick publish test

Build a source/wheel and check locally:

```
python -m pip install --upgrade build twine
python -m build
python -m twine check dist/*
```

Upload (example):

```
python -m twine upload dist/*
```
