Metadata-Version: 2.4
Name: areno
Version: 0.0.1.dev0
Summary: Extreme single-node scale-up infrastructure for LLM post-training.
Author: asystem
License: Apache-2.0
Project-URL: Homepage, https://github.com/asystem/areno
Project-URL: Repository, https://github.com/asystem/areno
Keywords: llm,post-training,rlhf,reinforcement-learning,cuda,inference
Classifier: Development Status :: 2 - Pre-Alpha
Classifier: Intended Audience :: Developers
Classifier: Intended Audience :: Science/Research
Classifier: License :: OSI Approved :: Apache Software License
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.10
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Requires-Python: >=3.10
Description-Content-Type: text/markdown

# areno

**Extreme single-node scale-up infrastructure for LLM post-training.**

> This is a name placeholder release. The project is not yet generally
> available on PyPI. Development happens at
> <https://github.com/asystem/areno>.

`areno` is a CUDA-native post-training and inference stack for local LLMs:
one workflow for RL/SFT training, tensor-parallel rollout, Hugging
Face-compatible checkpointing, and an OpenAI-compatible serving endpoint —
all on a single node.

Until the first general release, install from source following the
instructions in the [GitHub repository](https://github.com/asystem/areno).
