Metadata-Version: 2.4
Name: pyxlog
Version: 0.9.2
Classifier: Programming Language :: Rust
Classifier: Programming Language :: Python :: 3
Classifier: Operating System :: POSIX :: Linux
Requires-Dist: scipy ; extra == 'test'
Provides-Extra: test
Summary: GPU-accelerated XLOG probabilistic logic programming (PyO3 bindings)
License: MIT OR Apache-2.0
Requires-Python: >=3.8
Description-Content-Type: text/markdown; charset=UTF-8; variant=GFM
Project-URL: Documentation, https://github.com/BrainyBlaze/xlog/blob/main/docs/architecture/python-bindings.md
Project-URL: Homepage, https://github.com/BrainyBlaze/xlog
Project-URL: Issues, https://github.com/BrainyBlaze/xlog/issues
Project-URL: Repository, https://github.com/BrainyBlaze/xlog

# pyxlog

`pyxlog` provides Python bindings for the XLOG GPU-accelerated probabilistic
logic programming runtime.

The package is built from the `BrainyBlaze/xlog` repository and exposes the
native extension module together with the staged CUDA kernel artifacts needed by
the runtime.

At import time, `pyxlog` prefers packaged kernel artifacts under
`pyxlog/kernels/` and exports that path to `XLOG_CUBIN_DIR` automatically when
the wheel includes them. For source-tree validation, ad-hoc probe scripts, or
artifact runners that execute without the packaged kernel directory, set
`XLOG_CUBIN_DIR` explicitly to a directory containing `.cubin` or
`.portable.ptx` files before importing `pyxlog`.

Project documentation, setup instructions, and release notes live in the
repository root:

- https://github.com/BrainyBlaze/xlog

Use the root project README for installation requirements, CUDA expectations,
and end-to-end examples. For local source-tree installs, use the repository
helper from the root directory so the wheel is built for and installed into the
same Python interpreter your downstream project uses:

```bash
python scripts/install_pyxlog_for_python.py --python /usr/local/bin/python --user
```

