align: open post-training stack and Lean 4 prover research (code)

Snapshot of 2026-10-19:

code/: the working tree at the commit named in COMMIT.
align.bundle: a git bundle with the full history. Restore it with git clone align.bundle align.

Data is at https://huggingface.co/datasets/viai957/lean-prover-data.

The code is licensed Apache-2.0 (see code/LICENSE); third-party attributions are in code/NOTICE. The pre-registration is code/docs/PREREG_SOTA.md, OpenTimestamps-stamped under code/docs/evidence/prereg/.

Status: Phase 0, with no trained checkpoint yet.

Model revisions used: code/docs/evidence/model_revisions_deleted_2026-10-19.txt.

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support