TabPFN-Rel 0.0.6

Generated by Claude Code

TabPFNRel documents its arguments in Google-style docstrings. TabPFN-Rel requires relarena-core 0.0.6 or later; relarena-core 0.0.7 adds checks for incomplete validation windows and dropped test entities. The relational-prediction agent skill for Claude Code and Codex installs from this repository as a plugin; see Let a coding agent set it up.

Install with pip install "tabpfn-rel[api]==0.0.6" or pip install "tabpfn-rel[local]==0.0.6".

TabPFN-Rel 0.0.5

Generated by Claude Code

tabpfn-rel-client-latest and tabpfn-rel-local-latest are RelArena model aliases for the current client and local versions; results record the dated ID. The first local model is renamed from tabpfn-rel-local to tabpfn-rel-local-2026-08-15. Requires relarena-core 0.0.5.

Install with pip install "tabpfn-rel[api]==0.0.5" or pip install "tabpfn-rel[local]==0.0.5".

TabPFN-Rel 0.0.4

Generated by Codex

Use TabPFN 3.5 for the local and client defaults. The hosted backend uses simple text processing; the local backend uses native TF-IDF. Previous defaults remain available through client-2026-08-15 and local-2026-08-15.

Install with pip install "tabpfn-rel[api]==0.0.4" or pip install "tabpfn-rel[local]==0.0.4".

TabPFN-Rel 0.0.3

Generated by Codex

Require relarena-core 0.0.3 or newer. This brings the datasets compatibility constraint to standalone API and local installations. Update the package and runtime versions to 0.0.3.

Install with pip install "tabpfn-rel[api]==0.0.3" or pip install "tabpfn-rel[local]==0.0.3".

TabPFN-Rel 0.0.2

Generated by Codex

Release TabPFN-Rel from its standalone repository. Fit reusable PredictiveContext objects with TabPFNRel and pass explicit PredictiveQuery objects for prediction. Use the published relarena-core 0.0.2 interface and sparse-column handling.

Install with pip install "tabpfn-rel[api]==0.0.2" for the API backend or pip install "tabpfn-rel[local]==0.0.2" for the local backend.

The predictive interface has changed: fit a context, then supply a query to predict. The installed wheel passed 64 tests; three opt-in local checkpoint tests were skipped.

PyPI package