Metadata-Version: 2.4
Name: relarena
Version: 0.1.1
Summary: A unified, fair benchmarking framework for models on relational tasks on RelBench databases.
Keywords: benchmarking,machine-learning,relational-learning,relbench,tabular-data
Author: Prior Labs
License-Expression: Apache-2.0
License-File: LICENSE
License-File: NOTICE
License-File: src/relarena/models/VENDORED-LICENSES
Classifier: Development Status :: 3 - Alpha
Classifier: License :: OSI Approved :: Apache Software License
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Requires-Dist: relarena-core==0.0.3
Requires-Dist: relbench==2.1.2
Requires-Dist: numpy>=1.24
Requires-Dist: pandas>=2.3.3,<3.0
Requires-Dist: scikit-learn>=1.3
Requires-Dist: configspace>=1.0
Requires-Dist: pyyaml>=6.0
Requires-Dist: jsonschema>=4.0
Requires-Dist: torch
Requires-Dist: relarena[rdl] ; extra == 'graphsage'
Requires-Dist: kurversc==0.1.1 ; extra == 'kurversc'
Requires-Dist: bencheval ; extra == 'leaderboard'
Requires-Dist: lightgbm>=4.0,<4.7 ; extra == 'lightgbm'
Requires-Dist: bencheval[plot] ; extra == 'plots'
Requires-Dist: autorank ; extra == 'plots'
Requires-Dist: tabpfn>=8 ; extra == 'rdblearn'
Requires-Dist: relarena-core[dfs]==0.0.3 ; extra == 'rdblearn'
Requires-Dist: torch-geometric>=2.5 ; extra == 'rdl'
Requires-Dist: pytorch-frame>=0.2.3 ; extra == 'rdl'
Requires-Dist: sentence-transformers ; extra == 'rdl'
Requires-Dist: relarena[rdl] ; extra == 'relgnn'
Requires-Dist: relarena[rdl] ; extra == 'relgt'
Requires-Dist: einops>=0.8 ; extra == 'relgt'
Requires-Dist: h5py>=3.0 ; extra == 'relgt'
Requires-Dist: relational-transformer==1.8.0 ; extra == 'rt'
Requires-Dist: tabpfn-rel[api]>=0.0.2 ; extra == 'tabpfn-rel-api'
Requires-Dist: tabpfn-rel[local]>=0.0.2 ; extra == 'tabpfn-rel-local'
Requires-Python: >=3.11, <3.13
Project-URL: Homepage, https://github.com/PriorLabs/relarena
Project-URL: Repository, https://github.com/PriorLabs/relarena
Project-URL: Issues, https://github.com/PriorLabs/relarena/issues
Project-URL: Paper, https://arxiv.org/abs/2608.16319
Provides-Extra: graphsage
Provides-Extra: kurversc
Provides-Extra: leaderboard
Provides-Extra: lightgbm
Provides-Extra: plots
Provides-Extra: rdblearn
Provides-Extra: rdl
Provides-Extra: relgnn
Provides-Extra: relgt
Provides-Extra: rt
Provides-Extra: tabpfn-rel-api
Provides-Extra: tabpfn-rel-local
Description-Content-Type: text/markdown

# RelArena

RelArena benchmarks relational prediction models with shared temporal validation,
tuning and reporting. It includes baseline models and adapters for RelBench tasks.

```bash
pip install relarena
relarena --list
```

Install TabPFN-Rel for benchmarking with a local or hosted backend:

```bash
pip install "relarena[tabpfn-rel-local]"
# Or: pip install "relarena[tabpfn-rel-api]"
relarena --model tabpfn-rel-local --datasets rel-f1 --tasks driver-dnf --n-trials 1
```

This distribution depends on `relarena-core`. TabPFN-Rel is optional and can also
be installed directly as `tabpfn-rel[local]` or `tabpfn-rel[api]` for prediction on
user databases without the benchmark package.

TabPFN-Rel is developed in [its own repository](https://github.com/PriorLabs/tabpfn-rel). See the
[repository README](https://github.com/PriorLabs/relarena#readme) for model support,
installation extras, cache usage and benchmark commands, and the
[contributor guide](https://github.com/PriorLabs/relarena/blob/main/CONTRIBUTING.md)
for workspace development.
